A Longitudinal Analysis of Job Skills for Entry-Level Data Analysts

Trinity University Trinity University

Digital Commons @ Trinity Digital Commons @ Trinity

School of Business Faculty Research School of Business

Fall 2020

A Longitudinal Analysis of Job Skills for Entry-Level Data Analysts A Longitudinal Analysis of Job Skills for Entry-Level Data Analysts

Tianxi Dong

Trinity University

, tdong@trinity.edu

J. Triche

Follow this and additional works at: https://digitalcommons.trinity.edu/busadmin_faculty

Part of the Business Administration, Management, and Operations Commons

Repository Citation Repository Citation

Dong, T., & Triche, J. (2020). A longitudinal analysis of job skills for entry-level data analysts.

Journal of

Information Systems Education, 31

(4), 312-326.

This Article is brought to you for free and open access by the School of Business at Digital Commons @ Trinity. It

has been accepted for inclusion in School of Business Faculty Research by an authorized administrator of Digital

Commons @ Trinity. For more information, please contact jcostanz@trinity.edu.

Journal of

Information

Systems

Education

Volume 31

Issue 4

Fall 2020

A Longitudinal Analysis of Job Skills for Entry-Level

Data Analysts

Tianxi Dong and Jason Triche

Recommended Citation: Dong, T. & Triche, J. (2020). A Longitudinal Analysis of Job Skills for

Entry-Level Data Analysts. Journal of Information Systems Education, 31(4), 312-326.

Article Link: http://jise.org/Volume31/n4/JISEv31n4p312.html

Initial Submission: 27 September 2019

Accepted: 27 March 2020

Abstract Posted Online: 8 September 2020

Published: 10 December 2020

Full terms and conditions of access and use, archived papers, submission instructions, a search tool,

and much more can be found on the JISE website: http://jise.org

ISSN: 2574-3872 (Online) 1055-3096 (Print)

A Longitudinal Analysis of Job Skills for Entry-Level Data

Analysts

Tianxi Dong

School of Business

Trinity University

San Antonio, TX 78212, USA

tianxi.dong@trinity.edu

Jason Triche

University of Montana

Missoula, MT 59801, USA

jason.triche@umontana.edu

ABSTRACT

The explosive growth of the data analytics field has continued over the past decade with no signs of slowing down. Given the fast

pace of technology changes and the need for IT professionals to constantly keep up with the field, it is important to analyze the job

skills and knowledge required in the data analyst and business intelligence (BI) analyst job market. In this research, we examine

over 9,000 job postings for entry-level data analytics jobs over five years (2014-2018). Using a text mining approach and a custom

text mining dictionary, we identify a preliminary set of analytic competencies sought in practice. Further, the longitudinal data also

demonstrates how these key skills have evolved over time. We find that the three biggest trends include proficiency with Python,

Tableau, and R. We also find that an increasing number of jobs emphasize data visualization. Some skills, like Microsoft Access,

SAP, and Cognos, declined in popularity over the time frame studied. Using the results of the study, universities can make informed

curriculum decisions, and instructors can decide what skills to teach based on industry needs. Our custom text mining dictionary

can be added to the growing literature and assist other researchers in this space.

Keywords: Business analytics, Business intelligence, Careers, Employment skills, Job skills, Text processing

1. INTRODUCTION

The explosive growth of the data analytics field has continued

over the past decade with no signs of slowing down. According

to CIO.com, the data analytics field included two of the top five

hot and high-paying tech skills in 2018 (Scorsone, 2018).

LinkedIn also published the top seven skills in high demand for

2019, and two of these seven skills were in a data analytics-

related field (Bila, 2018). IBM predicts that the demand for data

analysts will increase by 28% by 2020 (Columbus, 2017).

Besides the fact that the number of jobs greatly outpaces the

number of qualified candidates, there are other concerns in the

field. One of the major concerns is defining what exactly data

analysts do. The data analyst and business intelligence (BI)

analyst fields vary greatly, and professionals have a hard time

understanding the skills needed. In one study, Bowne-Anderson

(2018) spoke to over 30 data analysts and BI analysts across a

wide array of industries and academic disciplines to find out

what their jobs entail. One highlight of the interview suggests

that the ability to communicate the results of analytics is more

important than the ability to use sophisticated deep learning

models. Other results suggest that specialization is becoming

more important and that data ethics need to come from within

data science itself as well as from legislation, grassroots

movements, and other stakeholders.

Given the fast pace of technological changes and the need

for IT professionals to constantly keep up with the field, it is

important to analyze the job skills and knowledge required in

the data analyst and BI analyst job market. Because the field

changes so rapidly, it is also just as important to understand

what data analyst and BI analyst skills are trending upward as

well as trending downward over time. Currently, only a few

studies (Deng, Li, and Galliers, 2016; Luo, 2016) have

attempted to address these concerns and have not fully explored

the changing trends. To address this gap, we seek to answer the

following questions:

1. What data analyst job skills and knowledge remained

steady from 2014-2018?

2. What data analyst job skills were popular in the past,

but are less attractive now?

Journal of Information Systems Education, Vol. 31(4) Fall 2020

312

3. W

hat data analyst job skills are gaining attention in the

current job market?

This study focuses on the job postings for entry-level hires

into the fields of data analyst and BI analyst. We define entry-

level hires as individuals who are completing their

undergraduate degree and wish to start a career in data and/or

BI analytics. Entry-level hires could also include individuals

who have an undergraduate degree in a field other than

business, math, or computer science or individuals who are

working in a non-analyst field and looking to change careers to

pursue a career in data and/or BI analytics. According to

GlassDoor, the data analyst job is one of the nine perfect jobs

for career changers (Moore, 2018). Our study explores the job

skills and knowledge for data analyst and BI analyst jobs on a

large scale. Using text analysis, we analyze approximately

9,000 entry-level job postings from Indeed.com in the data

analyst and BI analyst field from 2014 to 2018.

The results of this study have several practical

contributions. First, as undergraduates start to define their

career aspirations in their last years of college, these findings

can help focus their skills. Students can start choosing electives

that strengthen certain skills that are in hot demand in the

current data and BI analyst job market. Second, the results of

this research can also assist those individuals who want to

change their current career to an entry-level data or BI analyst

role. Those individuals can start pursuing professional training,

university certificates, or online courses that can teach the skills

that are in hot demand in this market. Third, these findings can

help undergraduate programs and professional training

companies focus their course offerings to align with the latest

skills in this rapidly changing job market. Lastly, the results of

this research can help companies retool and train their

employees on the latest data and BI analyst skill sets to stay

competitive in a data-driven business ecosystem.

2. LITERATURE REVIEW

2.1 Data Science, Data Analytics, and Business Intelligence

The concepts of data science, data analytics, and business

intelligence are not new. One of the founding disciplines,

statistics, has been around since the mid-1700s. Mathematicians

have been building models for centuries, and even basic

database concepts have been around since the 1960s.

Companies have been analyzing data to improve consumer

interaction, make production more efficient, and reduce cost

since the middle of the 20th century (Mills, Chudoba, and

Olsen, 2016). However, there has been explosive growth in

these fields over the past decade. Mills, Chudoba, and Olsen

(2016) have attributed this growth to three trends. The first

trend is enhanced technology infrastructure that can handle

terabytes of data in real time (Silva et al., 2014). The second

trend is the advancement of data storage and transformation.

The third trend is the expansion of analytical tools and

techniques (Chen, Chiang, and Storey, 2012; Davenport, Barth,

and Bean, 2012). These three trends have allowed the field to

expand from its established foundations.

The terms data science, data analytics, business

intelligence, and big data are used throughout the literature, but

it is helpful to define these terms. Data science is a field that

finds value in data and uses this value to create additional data

products (Loukides, 2011). Davenport and Patil (2012) define a

data scientist as a combination of a data hacker, analyst,

communicator, and trusted adviser.

Data analytics refers to the process of inspecting, cleaning,

transforming, and modeling data with the goal of supporting

decision-making (Lewis-Beck, 1995). Business analytics is a

subfield of data analytics that refers to problem recognition and

problem solving that happens within the context of business

situations (Holsapple, Lee-Post, and Pakath, 2014). In one

study, Turel and Kapoor (2016) examined the gap between

business school curricula and presumed industry needs in

business analytics. They identified business analytics courses

as those that cover topics such as “business intelligence, data

visualization, big data, and their roles in business strategy and

in improving business performance” (p. 99).

The Gartner Group introduced the term business

intelligence in 1989, describing a set of concepts and methods

to improve business decision making by using fact-based

support systems (Power, 2007). Expanding on this definition,

business intelligence is a broad category of applications,

technologies, architectures, and processes for gathering,

storing, accessing, and analyzing operational data to provide

business users with timely competitive information to enable

better insights for operational and strategic decision making

(Gupta, Goul, and Dinter, 2015).

The term big data is more than just a large amount of data;

it also includes the tools and procedures used to manipulate and

analyze the data (Burkholder, 1992). Big data is driving

changes in what types of data are being collected, how often

data are collected, and how much data are being collected

(Gardiner et al., 2018). Big data also has the ability to analyze

varied datasets with respect to variety (structured and

unstructured data), volume (amount of data), and velocity

(archival versus streaming data).

To help clarify the distinction between the terms data

science and data analytics, Aasheim et al. (2015) examined a

small sample of undergraduate programs in the U.S. Some

programs called themselves data science programs and some

called themselves data analytics programs. They found that

there are similarities between the two differently named

programs. These similarities include an increase in the number

of courses on statistics, data management, data mining, data

visualization, and other modeling techniques. They also found

several differences between the programs. Mainly, data science

programs were interdisciplinary in nature and required

additional math (at least through linear algebra), programming,

and statistics courses, whereas analytics programs covered data

warehousing and were mainly located in the business school.

These data science characteristics align with the definition of

data science as defined above by Loukides (2011) and

Davenport and Patil (2012).

As far as the current job market, ac

cording to a report from

Burning Glass Technologies, Business-Higher Education

Forum, and IBM, the fields of data-driven decision makers,

functional analysts, data analysts, and analytical managers all

had double-digit projected job growth over the next five years

(Miller and Hughes, 2017). Another report by PWC estimated

that by 2021, only 23% of educators stated that graduates would

have BI&A skills, while 69% of the employers prefer job

candidates with those skillsets (PWC, 2017).

Journal of Information Systems Education, Vol. 31(4) Fall 2020

313

2.2 Skills, Knowledge, and Abilities

Since we are analyzing each job posting in its entirety, there are

several different categories to examine. These categories

include skills, knowledge, and abilities. Skills are competencies

developed through training or experience, knowledge is the

theoretical or practical understanding of a subject, and abilities

are the talents involved in being able to do something (Lauby,

2013).

There are two similar studies that investigate the business

analytics job market. In one of the studies, Deng, Li, and

Galliers (2016) investigated the skills, knowledge, and abilities

that employers required for a business analytics role. In

addition, they analyzed the job postings by business degree

versus non-business degrees (i.e., computing, engineering,

statistics, or mathematics). Using Latent Semantic Analysis,

Deng, Li, and Galliers analyzed 71 unique job postings from

LinkedIn. They discovered that the postings were split fairly

evenly between business and non-business degrees and that the

top two industries looking for employees were healthcare and

information technology services. In addition to their findings,

the researchers mapped the required skills to Bloom’s

taxonomy (Bloom, 1956) to help institutions align course

design with student learning objectives.

The other similar study analyzed 1,216 job advertisements

from Indeed.com that contained the word “big data” in the job

title (Gardiner et al., 2018). Using the pile-sort methodology,

they found two major insights. The first insight was that a

number of job postings emphasized the design and development

of analytical systems. The second insight was that soft skills

still remain very important in job postings, even given the

demand for skills in new and emerging technologies. In

addition to these insights, the authors created a Big Data

Discipline Skills Conceptual Model. This model highlights and

provides insight into the complexity of the big data discipline.

It is worth noting that we found it difficult to differentiate

between skills and abilities. Given the definition that skills are

competencies and abilities are talents, it is hard to determine the

difference with static job postings. For example, is a job posting

asking for an applicant to generate data mappings a skill or an

ability? For our purposes, the category does not matter, as it is

mandated by the employer. Therefore, we followed Gardiner et

al. (2018) in combining the terms skills and abilities into one

term, which we will refer to as skills. We further break down

skills into two groups. The first group is higher-level general

domain skills (e.g., business intelligence and statistics), and the

second group is specific software skills (e.g., SAP and R). We

refer to knowledge as the academic background needed for the

job posting (i.e., high school, bachelor’s degree, master’s

degree, or doctoral degree).

2.3 Prior Relevant Studies

Another study closely related to our research is a study of the

requirements needed for entry-level analytics professionals.

Luo (2016) examined job ads posted on three major sites

(LinkedIn, Monster.com, and Indeed.com) over a six-month

period. After analyzing 924 job postings, the author concluded

that companies expected entry-level professionals to be able to

work in teams and work with databases and spreadsheets. The

study also found that data mining, optimization, and other

advanced analytical methods were rarely listed in postings for

entry-level positions.

A related study took a broader approach by examining

trends in required job skills for IT professionals from 1988 to

2003 (Gallivan, Truex, and Kvasny, 2004). The researchers

performed content analysis on 17 years of samples from the job

ads section of Computerworld and the Sunday classified job ads

section of a major metropolitan newspaper. Their goals were to

1) determine the most dramatic trends for IT positions, 2)

determine the most dramatic trends for required skills, and 3)

prove the forecasts offered by researchers in earlier studies were

accurate. They found that most of their results were consistent

with previous forecasts. For example, employers were asking

for an ever-increasing number and variety of skill sets from new

hires, and programming and software development skills

remained important. They also discovered that although

employers emphasized the need for well-rounded individuals

and strong soft skills, the job postings prioritized hard skills.

Using a web content data mining application, one study

(Aken et al., 2010) examined approximately a quarter-million

unique IT job descriptions. Using cluster analysis, they

condensed job postings from various job search engines into 20

different job skill clusters. At the time of publication, the top

five job skill clusters were IT managers, security specialists,

project analysts/managers, system administrators, and database

developers. Aasheim et al., (2012) presented an extension to

two previous studies (Aasheim, Li, and Williams, 2009;

Aasheim, Williams, and Butler, 2009) in which they examined

knowledge and skill requirements for entry-level IT employees.

Using a survey and a basic z-test, they discovered that the top

skills needed by employers were personal and interpersonal

skills, with honesty and integrity ranked most highly.

Text analyses of job postings ha

ve been performed in other

business- and information-related fields. In the field of

information science, one research project examined job

descriptions and advertisements for three data curation-focused

positions (Lyon et al., 2015). Their goals were to 1) discover

what skills were required for data science roles, 2) map data

science roles to current curriculum topics and course offerings,

and 3) develop new collaborations and partnerships in the data

science curriculum. Using job postings from Indeed.com from

January 2014 to April 2015, the researchers examined postings

in the fields of data librarian, data curator, and data archivist.

Among other findings, the research demonstrated that

employers were seeking data-savvy graduates who were work-

ready.

Another study in the field of marketing used content

analysis to examine 500 marketing job postings ranging from

entry-level all the way to senior-level positions (Schlee and

Harich, 2010). Not surprisingly, they found that there were

considerable differences in the skills and knowledge required

for all different levels of marketing jobs. They also discovered

that technical skills appeared to be much more important than

what had been documented in earlier marketing research.

In summary, there are several studies that examine the data

analytics job market. Our study differs from these previous

studies in several ways. First, we examine a longer time frame

(i.e., five years) compared to only several months (e.g., Luo,

2016; Gardiner et al., 2018). We also analyze about 9,000 job

postings compared to other studies that only analyzed about

1,000 job postings (e.g., Luo, 2016; Gardiner et al., 2018).

Lastly, we use several terms to search for job postings

compared to a specific search term like “big data” (Gardiner et

Journal of Information Systems Education, Vol. 31(4) Fall 2020

314

l., 2018), “BA positions” (Deng, Li, and Galliers, 2016), or

“data analyst” (Luo, 2016).

3. METHODOLOGY

In this study, we use a text mining approach to systematically

analyze job postings to identify the skills and knowledge

expected for entry-level data analyst positions. We follow the

established protocols in previous IS research (Mudambi and

Schuff, 2010; Wang, Kannan, and Ulmer, 2013) for data

collection and text mining analysis. Figure 1 provides an

overview of the data collection, processing, and major analyses

for this study.

Figure 1. Data Collection, Processing, and Analysis

3.1. Data Collection

3.1.1 Data source. We started by examining job postings on

Monster.com, Indeed.com, and LinkedIn.com. We discovered

that all three job posting sites shared mostly the same job

postings. Therefore, we decided to focus our search on just one

of the job posting websites to avoid duplication. In 2010,

Indeed.com passed Monster.com to become the highest-traffic

job website in the United States. Currently, Indeed.com boasts

that they have over 3 million companies with 250 million job

seekers (http://www.Indeed.com

Our first challenge was how to collect job postings over

past years because job websites do not keep historical data for

more than one year. Therefore, we used the Common Crawl

dataset to address this problem (http://commoncrawl.org/

Common Crawl is a non-profit organization that builds and

maintains an open repository of web crawl data that is, in

essence, a copy of the Internet. Common Crawl data contains

over 25 billion web pages (Batikas, Claussen, and Peukert,

2018) and is widely used in hundreds of research projects

(Batikas, Claussen, and Peukert, 2018; Cafarella et al., 2018).

Since we were only interested in the content from Indeed.com,

we only examined a very small fraction of the Common Crawl

corpus.

3.1.2 Search terms. A key component of any text analysis is to

first define the search terms to be used. We examined previous

studies (Sodhi and Son, 2009; Liberatore and Luo, 2013; Deng,

Li, and Galliers, 2016; Luo, 2016) to determine key search

terms. We used a wildcard search of business analy*, data

analy*, and business int*. These wildcard searches returned the

following results: data analyst, data analytics associate, data

analysis, business intelligence analyst, and business analytics.

The scope of this research is examining entry-level

positions. During our initial search process, we used the search

term “data scientist” (technically, we used a wildcard search of

data sc*). After examining a random sample of the job postings

returned, we discovered that these job postings required at

minimum a bachelor’s degree plus at least a few years of work

experience in the data analytics field. We considered these not

to be entry-level positions and out of the scope of our research.

Therefore, we excluded the terms data scientist, data science

associate, and data science analyst from our search terms. The

total number of job postings downloaded from Common Crawl

for each year from 2014 through 2018 that fit our search criteria

is listed in Table 1

Year

Number of Job Postings

2014

1722

2015

1955

2016

1735

2017

1888

2018

1863

Table 1. Number of Job Postings by Year

3.2. Data Processing

In order to accurately and automatically acquire job skill

content from large-scale texts, natural language processing

(NLP) techniques must be introduced (Jurafsky and Martin,

2008). According to research in computational linguistics

(Church, 1988) and our observations of the words describing

job skills in the job postings, the main job skill information

comes from nouns. Therefore, verbs, adjectives, and

prepositions could be safely removed without losing the main

information. We applied part-of-speech tagging to filter the

nouns for further analysis.

The data processing can be broken down into the following

steps:

1. Convert 9,163 job postings to individual words. These

job postings have a total of 4,499,672 words and 35,765

unique words.

2. Normalize/tokenize words. By applying lemmatization,

total words and unique words are reduced to 2,867,392

and 24,613, respectively.

3. Apply part-of-speech (POS) tagging and chunking.

There are 12,490 unique noun words and phrases

remaining.

4. Remove rare or common nouns and noun phrases. Any

nouns and noun phrases that appear in fewer than 1%

or more than 75% of the job postings are removed. The

number of unique noun words and phrases is further

reduced to 6,067. With 6,067 unique nouns and noun

phrases left, it is feasible for us to manually analyze the

key words that are relevant to data analyst job skills.

3.2.1 Part-of-speech tagging (POS). Part-of-speech tagging is

a process of assigning morphological tags or categories

Journal of Information Systems Education, Vol. 31(4) Fall 2020

315

(classes) to each token (Voutilainen, 2003). Commonly, in the

context of NLP, a token is preferred, rather than a word, because

each token has a unique meaning. After POS tagging is applied,

the tokens in a sentence are marked into proper classes. In this

research, the Stanford Log-linear Part-Of-Speech Tagger,

which was developed by the NLP group at Stanford University,

has been adopted. The Stanford Log-linear Part-Of-Speech

Tagger is considered the most widely used POS tagger, as this

tagging system can achieve more than 95% accuracy in English

tagging (Toutanova and Manning, 2000).

3.2.2 Chunking. It is not uncommon that single noun tokens

can fail to represent some job skills, such as interpersonal skills

(adj + noun) or Microsoft Office (noun + noun). In NLP, these

short phrases are identified as noun phrases (NP). It is more

appropriate to consider these NPs combined as a noun token,

rather than separate tokens. Chunking addresses these issues.

Chunking is a process of extracting phrases from text data.

Therefore, in addition to using part-of-speech tagging,

chunking is also applied to extract NP. Tokens that follow the

grammatical patterns described below are extracted for further

analysis:

1. noun, for example, Python, R, Excel, Tableau

2. noun + noun, for example, Microsoft Office,

management software, data visualization

3. gerund + noun, for example, programming skills,

writing skills

4. adjective + noun, for example, interpersonal skills,

analytic capability

3.2.3 Dictionary development. A key step in text mining

involves creating or using an existing dictionary of relevant

words in order to categorize and classify the search results.

Because there is sparse research in this area and the field is

constantly evolving, we did not have an exhaustive dictionary

to reference. Therefore, we built our own dictionary. There are

two ways to build a dictionary from scratch – a theoretical

approach and an empirical approach (Luo, 2016).

The theoretical approach involves using existing terms in

relevant literature. We started by examining literature that has

used a text mining approach to identify job skills in business

analytics (Deng, Li, and Galliers, 2016) and operational

research skills (Sodhi and Son, 2009). Starting with a list of 367

words, two expert coders in the field of data analytics agreed on

a final list of 149 words that are relevant to the data analytics

field. We started our dictionary with these 149 words. The

Kappa interrater reliability was 0.984, which is above the well-

established mark of .70 according to Landis and Koch (1977)

and Bowers and Courtright (1984). Coding disagreements were

discussed, and eventually a consensus was reached.

Although these 149 words were a good starting point, at

first glance they did not fully reflect the current job market of

the data analytics field. For example, the programming

language Python was not included in the original list although

it is a very popular programming language in the field.

Therefore, we used an empirical approach to update our

dictionary. An empirical approach involves deriving words

from a large sample of representative job ads (Sodhi and Son,

2009). Using the original corpus of job postings from

Indeed.com, we identified 6,067 single words. A vast majority

of these words were not relevant since they included every word

in a job posting. For example, the words “position,” “user,” and

“education” were not useful in our dictionary. The same two

coders agreed on a final list of 40 words to add to the dictionary.

New additions to the dictionary included words like “Python,”

“R,” and “Pentaho,” which are all relevant to the data analytics

field. The Kappa interrater reliability was 0.956, new coding

disagreements were discussed, and a consensus was reached.

Finally, similar concepts were merged and grouped into three

primary categories: general domain skills, software skills, and

knowledge. This process led to a dictionary containing three

primary categories with 14 sub-categories and 186 keywords

and phrases. The final dictionary is reported in Appendix A.

3.3 Statistical Analyses

In order to answer our research questions to determine if there

were significant differences in the skillsets between 2014 and

2018, we tested the difference between two population

proportions. The test statistic for measuring the difference

between two population proportions is:

where,

Longitudinal studies normally use some type of regression.

In our case, we did not have any variance within the year;

consequently, we could not get a standard error. Therefore,

regression would not work with our dataset. We also explored

a non-parametric test (i.e., MannKendall) to analyze the trend

each year. The MannKendall test examines the trend year over

year to determine if the trend is significant over the time period.

However, the minimum number of recommended

measurements for the MannKendall is at least 8-10 data points

(Khambhammettu, 2005). In our case, we only had five years

of data, resulting in only five measurements per job skill item.

This would result in inaccurate or biased results using a

MannKendall test. Therefore, the proportion test was the most

appropriate analytical method for our dataset.

4. RESULTS

This section displays the results of our analysis. Due to the large

number of keywords analyzed, only significant trends and the

most popular general domain skills, software skills, and

knowledge areas are displayed in this section. All results

(including non-statistically significant trends or less popular

skills and knowledge) are reported in Appendix B.

4.1 Analysis by U.S. State

First, we analyzed the results by U.S. state. The top five states

in 2018 with the highest number of data analyst jobs, in order,

were Virginia, Texas, California, New York, and Illinois. This

is not surprising given that three of the top five states in this list

are the largest states by population. Because the largest states

Journal of Information Systems Education, Vol. 31(4) Fall 2020

316

y population would have the largest number of job postings,

we standardized by dividing the total number of job postings by

the number of businesses in that state. We collected the number

of businesses in each state using the NAICS Association

website (NAICS, 2019). The top five states with the

standardized job postings, in order, were District of Columbia

(not a U.S. state), Virginia, Vermont, Delaware, and

Massachusetts. Figure 2 illustrates a heat map for the number

of standardized data analytics jobs in 2018.

Figure 2. Average Job Postings by State

Table 2 displays the percentage of job postings (job

postings in that state/job postings in the U.S.) by state for only

states that have a significant difference between 2014 and 2018.

States

2014

2015

2016

2017

2018

Virginia

9.36%

7.92%

11.92%

13.80%

18.00%***

Texas

7.02%

9.06%

8.16%

6.48%

10.20%***

New York

6.30%

7.44%

6.08%

5.10%

9.00%**

Massachusetts

2.52%

3.90%

4.48%

3.36%

4.08%**

Ohio

3.96%

3.30%

2.00%

2.58%

2.76%*

Washington

4.23%

2.34%

2.88%

2.40%

2.40%**

Maryland

3.24%

2.58%

3.84%

2.58%

2.28%*

Georgia

3.87%

3.78%

3.12%

1.80%

2.04%***

Table 2. Distribution of Data Analyst Job by U.S. State

*p < 0.05, **p < 0.01, ***p < 0.001

4.2 Evolution of Knowledge Required

Table 3 displays the percentages of job postings that required

bachelor’s, master’s, MBA, or Ph.D. degrees. Other kinds of

academic degrees (e.g., associate degree and high school

diploma) are not reported, as the percentages are below 1%,

which means few jobs required less than a bachelor’s degree

4.3 Evolution of General Domain Skills

Table 4 displays the percentages of job postings that required

general domain skills by year.

Knowledge Required

2014

2015

2016

2017

2018

Bachelor

60.60%

60.70%

62.30%

64.80%

70.30%***

Master

12.90%

12.10%

14.30%

15.90%

15.30%*

MBA

6.00%

4.10%

5.00%

4.60%*

Ph.D.

3.70%

3.80%

4.40%

5.60%**

Table 3. Evolution of Academic Requirements for Data Analysts

*p < 0.05, **p < 0.01, ***p < 0.001

Categories

2014

2015

2016

2017

2018

General Analytics

69.30%

69.70%

69.00%

71.60%

69.90%

General Statistics

22.30%

19.80%

23.00%

29.10%

28.20%***

Modeling

17.20%

15.90%

17.90%

20.30%

21.00%**

Model Development

21.30%

21.10%

22.80%

24.70%

25.70%***

Data Management

40.00%

41.20%

44.50%

50.30%

49.80%***

Table 4. Evolution of General Domain Skills for Data Analysts

*p < 0.05, **p < 0.01, ***p < 0.001

Journal of Information Systems Education, Vol. 31(4) Fall 2020

317

Software Skills

2014

2015

2016

2017

2018

Database System

48.00%

48.80%

52.40%

58.10%

58.90%***

SQL Server

10.40%

12.10%

13.40%

15.20%

17.80%***

Oracle

15.00%

14.00%

13.40%

14.40%

15.50%

Microsoft Access

13.60%

11.60%

11.30%

11.60%

10.40%**

NoSQL

0.70%

1.00%

1.60%

1.50%

1.60%**

DBMS

0.30%

0.60%

0.90%

0.90%*

Personal Productivity

30.70%

29.70%

32.60%

32.10%

28.60%

Microsoft Office

18.40%

16.30%

20.60%

18.50%

17.20%

Microsoft PowerPoint

12.10%

13.70%

14.20%

13.90%

Business Intelligence

11.30%

12.60%

17.60%

21.80%

23.20%***

Tableau

5.30%

8.30%

13.20%

18.20%

18.90%***

Cognos

7.50%

5.30%

5.80%

6.00%*

Power BI

0.10%

1.00%

1.60%

1.90%***

Programming Language

16.00%

16.40%

20.20%

21.90%

22.50%***

Python

2.30%

3.30%

6.60%

9.10%

10.70%***

Pig

0.30%

0.50%

0.70%

0.60%

1.00%**

Enterprise System

16.00%

15.80%

16.20%

18.50%

20.30%***

SAP

10.40%

9.00%

8.00%

8.40%

8.30%*

Hadoop

2.00%

2.40%

4.00%

4.50%

5.10%***

Salesforce

3.40%

4.50%

2.80%

4.50%

4.90%*

Azure

0.20%

0.00%

0.80%

1.30%

3.60%***

Hive

0.70%

1.10%

1.80%

2.10%***

Google Analytics

0.50%

1.50%

1.00%

1.20%*

Statistical Package

10.10%

9.50%

12.40%

15.40%

16.10%***

4.50%

4.40%

7.00%

9.60%

11.80%***

SAS

6.60%

8.00%

8.80%

9.70%***

SPSS

1.60%

1.90%

2.40%

3.70%

2.50%*

Table 5. Evolution of Software Skills for Data Analysts

*p < 0.05, **p < 0.01, ***p < 0.001

Categories are underlined; subcategories are indented. Bold (italicized) text indicates a positive (negative) trend between

2014 and 2018.

Journal of Information Systems Education, Vol. 31(4) Fall 2020

318

4.4 Evolution of Software Skills

Table 5 displays the percentages of job postings that required

software skills by year. The software skills are sorted by the

most popular software skills in 2018 within each category. Only

significant trends (i.e., those software skills that significantly

increased or decreased from 2014 to 2018) or software that

appeared in at least 10% of the job postings are displayed in

Table 5. Categories are underlined and subcategories are

indented. Bolded (italicized) software skills indicate a

significant increase (decrease) between 2014 and 2018. All

results are presented in Appendix B.

5. DISCUSSION

5.1 Discussion by Research Question

We will discuss the results in terms of the research questions.

Although the U.S. state analysis was not a research question, it

is important to know how data analyst jobs are distributed by

U.S. state. According to Table 2, Virginia increased from

9.36% of U.S. job postings in 2014 to 18% of U.S. job postings

in 2018. Virginia is the leading data center market in the U.S.

and has the 3rd-highest concentration of high-tech workers in

the nation. Virginia is “preparing for future growth for IT

companies through its top-ranked higher education system to

build a pipeline of technology talent” (Key Industries, n.d.). The

number of job postings in Texas increased gradually each year

as well. Several data centers were established in Texas recently,

such as Microsoft and RackSpace. These data centers are

developing rapidly, contributing to the increasing job demand

in Texas (Mosbrucker, 2018). It is also interesting to note that

several states showed declining percentages of job postings

from 2014 to 2018. For example, Ohio, Washington, Maryland,

and Georgia all show statistically significant decline over time,

although not a large number in magnitude.

The first research question is, “What data analyst job skills

and knowledge have remained steady from 2014 to

2018?” According to Table 4, general statistics has remained a

steady and highly desired skill over the time period studied. In

terms of software skills (see Table 5), Personal Productivity

Software (e.g., Visio, JIRA), Microsoft Office (not including

Access), and Oracle have remained steady and highly desired

during the time frame of the study. Other software or languages

that has remained steady, although not appearing in a large

percentage of job postings, include XML, Teradata, DB2,

MySQL, Linux, Visual Basic, and HTML. These general

domain skills and software skills have been steady for the past

several years and have been documented in other studies (e.g.,

Gallivan, Truex, and Kvasny, 2004; Luo, 2016).

The second research question is, “What data analyst job

skills were popular in the past, but are less attractive

now?” None of the general domain skills (see Table 4) showed

any decline from 2014-2018. In terms of software skills (see

Table 5), only Microsoft Access (p < 0.01), Cognos (p < 0.05),

and SAP (p < 0.05) showed statistically significant decline. In

terms of Microsoft Access, this may be due to direct

competition from other growing, open-source database

software like MySQL. There is some anecdotal evidence that

Microsoft Access customer support threads have been declining

(Microsoft, 2017). Cognos is an IBM business intelligence suite

that provides a toolset for reporting, analytics, scorecarding,

and monitoring of events and metrics. Although showing

decline, the suite is still solid at 6.00% of job postings in 2018.

SAP is an enterprise-wide software that helps manage

operations and customers. SAP also has BI software including

the BI suite, SAP Lumira, Hana, and Crystal Reports. Again,

although SAP has shown a statistically significant decrease

from 2014 to 2018, it is still a sought-after software skill at 8.3%

of job postings in 2018. In summary, three software skills

showed statistically significant decline, but these three software

packages are still desired skills in the market.

The third research question is, “What data analyst job skills

are gaining attention in the current job market?” There are

numerous upward-trending general domain skills and software

skills over 2014-2018. First, we will examine the upward-

trending skills that are in the top 25% of all job postings. The

percentages indicate the number of job postings requiring that

skill in 2018. General statistics (28%, p < 0.001), modeling

(21%, p < 0.01), model development (26%, p < 0.001), data

management (50%, p < 0.001), database systems (59%,

p < 0.001), BI (23%, p < 0.001), programming languages (23%,

p < 0.001), and enterprise systems (21%, p < 0.001) all

increased significantly from 2014 to 2018 and are highly

desired skills. In terms of software skills or languages, SQL

server (18%, p < 0.001), Tableau (19%, p < 0.001), statistical

packages (16%, p < 0.001), SAS (10%, p < 0.001), R (12%, p

< 0.001), and Python (11%, p < 0.001) are all in the top quartile

in terms of job postings and show a statistically significant

increase.

The next set of general domain skills and software packages

are increasing in demand over time but represent the next

quartile (top 50% of job postings). These software packages

include SPSS (3%, p < 0.05), Hive (2%, p < 0.001), Salesforce

(5%, p < 0.05), Hadoop (5%, p < 0.001), and Microsoft Azure

(4%, p < 0.001). Although demand for these skills grew during

this time frame, they do not represent the top quartile in terms

of the number of entry-level job postings asking for that

skillset.

Lastly, the skills that grew between 2014 and 2018 but

represent the lower 50% of the total entry-level job postings

include NoSQL (1.6%, p < 0.01), Microsoft Power BI (1.9%, p

< 0.001), Apache Pig (1%, p < 0.01), and Google Analytics

(1.2%, p < 0.05). An interesting observation is the lack of job

postings that mention NoSQL. NoSQL is a non-relational

database that is scaled horizontally and means “not only SQL.”

Only a small fraction of job postings mentioned NoSQL, which

indicates that NoSQL has not increased in popularity as

previously predicted (Pal, 2016). This is useful information for

instructors of database courses. If time is limited, instructors

should focus on relational databases instead of NoSQL since it

is not highly demanded in the industry.

There are several other software packages that are not

trending up or down (i.e, remained steady from 2014-2018) but

only represent a very small fraction of entry-level job postings.

For example, MongoDB (0.3%), Apache HBase (0.6%),

Apache Cassandra (0.2%), Pentaho (0.3%), JavaScript

visualization library – D3 (0.2%), STATA (0.5%), Ruby

(0.3%), and IBM Watson (0.2%) only appear in a small fraction

of job postings. Some of these software programs are taught in

database, analytics, and BI courses and are widely known in the

industry. However, the results of this research demonstrate that

they are not a widely needed skillset for entry-level data

analytics jobs. Therefore, given the time constraints of a course,

Journal of Information Systems Education, Vol. 31(4) Fall 2020

319

nstructors can leave these software packages out of the lesson

plan.

Lastly, the top three software and languages that grew the

fastest and are mentioned in at least 5% of the job postings in

2018, in order, include Python (11%), Tableau (19%), and R

(12%). These are widely used software and languages in

industry and are taught across the world in statistics, computer

science, and business intelligence courses. As of 2019, these are

highly sought-after skills and should remain part of the

curriculum in training programs and universities.

Other observations from the analysis include:

1. An increasing number of jobs require candidates to

have some programming skills, such as Python and R.

2. An increasing number of jobs emphasize data

visualization, which requires proficiency with software

like Tableau.

3. A large number of jobs require at least a bachelor's

degree (70%).

For data analysts, being familiar with interpreted

programming languages like Python and R is a prerequisite, as

programming is an inherent part of data processing. Data

visualization is the graphical representation of information and

data. By using visual elements like charts, graphs, and maps,

data visualization tools provide an accessible way to see and

understand trends, outliers, and patterns in data.

5.2 Contribution

This research provides a practical contribution for several

different audiences. First, universities can use these results to

help inform curriculum decisions. For example, the increasing

need for programming languages, data visualization, and

database skills is imperative. Second, instructors in an analytics

course can use these results to decide what technical and

software skills should be taught. As mentioned before, if time

is constrained, instructors could drop tools like Pentaho and

MongoDB. Third, this research is useful for companies that

provide business technology training. These training centers

can retool their offerings to match industry trends in skillsets

and software skills. Lastly, this research can help individuals

who are looking to change careers to data analytics. By

understanding the trending general domain skills and software

skills, individuals can start learning the most desired skills in

order to transition careers.

This research also provides some theoretical contributions.

First, as part of this research, we built a data dictionary that

other researchers and practitioners can use in the data analytics

text mining arena. We started with an established dictionary

from Deng, Li, and Galliers (2016) and Sodhi and Son (2009).

We then used an empirical approach to identify words from job

postings that were missing from the original list. This

comprehensive dictionary (see Appendix A) was built and

utilized for this research. Second, this research can help build

the literature in this area. To our knowledge, there are only a

few previously published articles that examine the data

analytics field using job postings. Our research has a larger

corpus and is more current.

5.3 Limitations

There are limitations to this research which should be noted.

First, the number of job postings listed is not comprehensive.

The job postings only include a sample of all the job postings

for that year. This is a limitation of the Common Crawl source.

As previously mentioned, Indeed.com does not store historical

data, so we used Common Crawl to retrieve the data from years

past. There could have been days when Common Crawl did not

archive the Indeed.com website, and during that time, a job

posting might have appeared and then disappeared. We can

make assertions about only a sample of job postings, not the

entire corpus of job postings.

Our second limitation is our statistical method to measure

trends over time. Since there is only one data point per year, we

could not calculate a variance or standard error. This prevented

us from using a longitudinal analysis using the five years as a

trend. Therefore, our trend analysis only covered two points in

time – 2014 and 2018. We did not capture a trend if a skill had

the same percentage in 2014 and 2018 (i.e., resulting in no

statistical significance in the proportions test) but either spiked

or plummeted in the years between. However, we examined

each item to verify whether this scenario existed, and we could

not find any item that had a significant rise or fall of the trend

between 2014 and 2018 with 2014 and 2018 having

approximately the same value. There is another possible

scenario in which the proportion test may indicate a positive

increase, but the overall trend is declining in the last few years.

For example, one item could start at 12% in 2014, spike to 20%

in 2015, then decline steadily over the next 3 years to 15% by

2018. Obviously, this is a declining trend and would be a fading

trend in the analytics space, but our proportion test would show

a significant increase between 2014 and 2018. We checked the

entire dataset and could not find an example of this trend.

Therefore, we are confident that the proportion test adequately

answers our research questions.

Our third limitation is that we only examined entry-level

positions and only general domain skills and software skills. A

comprehensive study would have examined all levels and

included soft skills. We decided not to include these analyses in

this research due to sample size limitations. These analyses

would be included in future research.

Our last limitation is that we examined job postings written

by employers themselves. Each company may have a different

idea of what a skillset is for their employees. Therefore, there

could be variance in vague general domain skills like general

statistics. However, this limitation is also a contribution of our

research since we summarized over 9,000 job postings over 5

years. Our summarized results show that general domain skills

like general statistics are still widely used in job postings.

6. CO

NCLUSION

In summary, this research examined job postings from 2014 to

2018 for entry-level data analytics jobs from Indeed.com. Using

a custom data dictionary built from previous literature and

empirical data, we employed a text mining approach to identify

word frequencies in the job postings. Using a difference of

proportions, we identified general domain skills and software

skills that trended over time. General analytics, general

statistics, modeling, and data management all appear in at least

20% of the job postings in our sample, with general analytics

Journal of Information Systems Education, Vol. 31(4) Fall 2020

320

ppearing in almost 70% of job postings. We find that Python,

Tableau, and R software skills are in high demand, and

Microsoft Access, Cognos, and SAP are in slow decline. We

also analyzed job postings by state and education level. We find

that at least a bachelor’s degree is required in 70% of the job

postings in our sample. Using the results of the study,

universities can make better-informed curriculum decisions,

and instructors can decide what skills to teach based on industry

needs. Our custom text mining dictionary can be added to the

growing literature and assist other researchers in this space. We

also identified limitations to this research and ideas for future

work.

7. ACKNOWLEDGEMENTS

Our thanks to the Journal of Information Systems Education

editor and reviewers for their advice and suggestions. We

would like to thank Dr. Robert Scherer whose comments and

suggestions helped improve and clarify this manuscript. This

research was supported by the Faculty Research Start-up Fund

and the Summer Research Stipend Program for 2018 of Trinity

University.

8. REFERENCES

Aasheim, C., Li, L., & Williams, S. (2009). Knowledge and

Skill Requirements for Information Technology Graduates:

A Comparison of Industry and Academia. Journal of

Information Systems Education, 20(3), 349-356.

Aasheim, C., Shropshire, J., Li, L., & Kadlec, C. (2012).

Knowledge and Skill Requirements for Entry-Level IT

Workers: A Longitudinal Study. Journal of Information

Systems Education, 23(2), 193.

Aasheim, C., Williams, S., & Butler, E. S. (2009). Knowledge

and Skill Requirements for IT Graduates: A Survey of IT

Workers and Managers. Journal of Computer Information

Systems, 42, 690-699.

Aasheim, C., Williams, S., Rutner, P., & Gardiner, A. (2015).

Data Analytics Vs. Data Science: A Study of Similarities and

Differences in Undergraduate Programs Based on Course

Descriptions. Journal of Information Systems Education,

26(2), 103-115.

Aken, A., Litecky, C., Ahmad, A., & Nelson, J. (2010). Mining

for Computing Jobs. IEEE Software, 27(1), 78-85.

Batikas, M., Claussen, J., & Peukert, C. (2018). Follow the

Money: Online Piracy and Self-Regulation in the

Advertising Industry (SSRN Scholarly Paper ID 3140358).

Social Science Research Network.

https://papers.ssrn.com/abstract=3140358

Bila, A. (2018). 7 Top Tech Skills In High Demand For 2019,

LinkedIn. Retrieved September 9, 2019, from

https://www.linkedin.com/pulse/7-top-tech-skills-high-

demand-2019-anna-bila.

Bloom, B. S. (1956). Taxonomy of Educational Objectives:

Handbook 1: Cognitive Domain. New York: David McKay.

Bowers, J. W. & Courtright, J. A. (1984). Communication

Research Methods. Glenview, Illinois: Scott, Foresman.

Bowne-Anderson, H. (2018). What Data Scientists Really Do,

According to 35 Data Scientists. Harvard Business Review.

Retrieved September 1, 2019, from

https://hbr.org/2018/08/what-data-scientists-really-do-

according-to-35-data-scientists.

Burkholder L. (1992). Philosophy and the Computer. Boulder,

Colorado: Westview Press.

Chen, H., Chiang, R. H., & Storey, V. C. (2012). Business

Intelligence and Analytics: From Big Data to Big Impact.

MIS Quarterly, 36(4), 1165-1188.

Church, K. W. (1988). A Stochastic Parts Program and Noun

Phrase Parser for Unrestricted Text. Second Conference on

Applied Natural Language Processing, 136-143.

Cafarella, M., Halevy, A., Lee, H., Madhavan, J., Yu, C., Wang,

D. Z., & Wu, E. (2018). Ten Years of Webtables.

Proceedings of the VLDB Endowment, 11(12), 2140-2149.

Columbus, L. (2017). IBM Predicts Demand for Data Scientists

Will Soar 28% By 2020. Forbes. Retrieved September 19,

2019, from

https://www.forbes.com/sites/louiscolumbus/2017/05/13/ib

m-predicts-demand-for-data-scientists-will-soar-28-by-

2020/#670c4c217e3b.

Davenport, T. H., Barth, P., & Bean, R. (2012). How Big Data

Is Different. MIT Sloan Management Review, 54(1).

Davenport, T. H. & Patil, D. (2012). Data Scientist. Harvard

Business Review, 90(5), 70-76.

Deng, X. N., Li, Y., & Galliers, R. D. (2016). Business

Analytics Education: A Latent Semantic Analysis of Skills,

Knowledge and Abilities Required for Business Versus Non-

Business Graduates. Proceedings from Thirty Seventh

International Conference on Information Systems, Dublin.

Gallivan, M. J., Truex III, D. P., & Kvasny, L. (2004). Changing

Patterns in IT Skill Sets 1988-2003: A Content Analysis of

Classified Advertising. ACM SIGMIS Database: the

DATABASE for Advances in Information Systems, 35(3), 64-

87.

Gardiner, A., Aasheim, C., Rutner, P., & Williams, S. (2018).

Skill Requirements in Big Data: A Content Analysis of Job

Advertisements. Journal of Computer Information Systems,

58(4), 374-384.

Gupta, B., Goul, M., & Dinter, B. (2015). Business Intelligence

and Big Data in Higher Education: Status of a Multi-Year

Model Curriculum Development Effort for Business School

Undergraduates, MS Graduates, and MBAs.

Communications of the Association for Information Systems,

36(23), 449-476.

Holsapple, C., Lee-Post, A., & Pakath, R. (2014). A Unified

Foundation for Business Analytics. Decision Support

Systems, 64, 130-141.

Jurafsky, D. & Martin, J. H. (2008). Speech and Language

Processing: An Introduction to Natural Language

Processing, Computational Linguistics, and Speech

Recognition. Prentice Hall Series in Artificial Intelligence,

1-1024.

Key Industries. (n.d.). Virginia Economic Development

Partnership, Retrieved September 1, 2019, from

www.vedp.org/key-industries

Khambhammettu, P. (2005). Mann-Kendall Analysis,

HydroGeoLogic Inc. OU-1 Annual Groundwater Monitoring

Report–Former Fort Ord, California.

Journal of Information Systems Education, Vol. 31(4) Fall 2020

321

andis, J. R. and Koch, G. G. (1977). The Measurement of

Observer Agreement for Categorical Data. Biometrics,

33(1), 159-174.

Lauby S. (2013). The Difference between Knowledge, Skills

and Abilities. HR Bartender Retrieved from

https://www.hrbartender.com/2013/recruiting/the-

difference-between-knowledge-skills-and-abilities/.

Lewis-Beck, M. S. (1995). Data Analysis: An Introduction,

Sage Publications, Inc.

Liberatore, M. & Luo, W. (2013). ASP, The Art and Science of

Practice: A Comparison of Technical and Soft Skill

Requirements for Analytics and OR Professionals.

Interfaces, 43(2), 194-197.

Loukides, M. (2011). What is Data Science? O’Reilly Media,

Inc. Retrieved August 22, 2019, from

https://www.oreilly.com/radar/what-is-data-science/

Luo, W. (2016). Responsibility and Skills Requirements for

Entry Level Analytics Professionals. Journal of

Organizational and End User Computing, 28(4), 1-14.

Lyon, L., Mattern, E., Acker, A., & Langmead, A. (2015).

Applying Translational Principles to Data Science

Curriculum Development. iPRES 2015.

Microsoft Community Support Pages. (2017). “Decline in use

of Access Database?” Retrieved April 29, 2019, from

https://answers.microsoft.com/en-

us/msoffice/forum/all/decline-in-use-of-access-

database/3407c710-7b98-4abe-835c-cb7c8b533641.

Miller, S. & Hughes, D. (2017). The Quant Crunch: How the

Demand for Data Science Skills is Disrupting the Job

Market. Retrieved from

https://www.bhef.com/publications/quant-crunch-how-

demand-data-science-skills-disrupting-job-market.

Mills, R. J., Chudoba, K. M., & Olsen, D. H. (2016). IS

Programs Responding to Industry Demands for Data

Scientists: A Comparison Between 2011-2016. Journal of

Information Systems Education, 27(2), 131.

Moore, E. (2018). 9 Perfect Jobs for Career Changers.

Glassdoor. Retrieved July 19, 2019, from

https://www.glassdoor.com/blog/jobs-for-career-changers/

Mosbrucker, K. (2018). Microsoft Plans for More Area Data

Centers Worth $62M, Bizjournals.com, Retrieved July 19,

2019, from

www.bizjournals.com/sanantonio/news/2018/09/07/microso

ft-plans-for-more-area-data-centers-worth.html.

Mudambi, S. M., & Schuff, D. (2010). What Makes A Helpful

Review? A Study of Customer Reviews on Amazon.com.

MIS Quarterly, 34(1), 185-200.

NAICS Association. (2019). Retrieved August 8, 2019, from

https://www.naics.com/business-lists/counts-by-state/

Pal, K. (2016). Why the World is Moving toward NoSQL

Databases?” Techopodia. Retrieved June 7, 2016, from

https://www.techopedia.com/2/32000/trends/big-data/why-

the-world-is-moving-toward-nosql-databases.

Power, D. J. (2007). A Brief History of Decision Support

Systems. Retrieved from DSSResources.com

PWC. (2017). Investing in America’s Data Science and

Analytics Talent. Retrieved from

http://pwc.com/us/dsa-

skills.

Schlee, R. & Harich, K. (2010). Knowledge and Skill

Requirements for Marketing Jobs in the 21

Century.

Journal of Marketing Education, 32(3), 341-352.

Scorsone, G. (2018). 5 Hot and High-Paying Tech Skills for

2018. Retrieved July 19, 2019, from

https://www.idginsiderpro.com/article/3269251/5-hot-and-

high-paying-tech-skills-for-2018.html.

Silva, Y. N., Dietrich, S. W., Reed, J. M., & Tsosie, L. M.

(2014). Integrating Big Data into the Computing Curricula.

Proceedings of the 45th ACM Technical Symposium on

Computer Science Education.

Sodhi, M. & Son, B. G. (2009). Content Analysis of OR Job

Advertisements to Infer Required Skills. Journal of the

Operational Research Society, 61(9), 1315-1327.

Toutanova, K. & Manning, C. D. (2000). Enriching the

Knowledge Sources Used in a Maximum Entropy Part-Of-

Speech Tagger. Proceedings of the 2000 Joint SIGDAT

Conference on Empirical Methods in Natural Language

Processing and Very Large Corpora: Held in Conjunction

with the 38th Annual Meeting of the Association for

Computational Linguistics, 13, 63-70.

Turel, O. & Kapoor, B. (2016). A Business Analytics Maturity

Perspective on The Gap Between Business Schools and

Presumed Industry Needs. Communications of the

Association for Information Systems, 39(6), 96-109.

Voutilainen, A. (2003). Part-Of-Speech Tagging. The Oxford

Handbook of Computational Linguistics. 219-232.

Wang, T., Kannan, K. N., & Ulmer, J. R. (2013). The

Association between the Disclosure and The Realization of

Information Security Risk Factors. Information Systems

Research, 24(2), 201-218.

AUTHOR BIOGRAPHIES

Tianxi Dong is an assistant professor of business analytics and

technology at Trinity University.

She earned her B.S. and M.S. in

management information systems

(MIS) from Shanghai University

of Finance and Economics. She

received her Ph.D. in MIS from the

Rawls College of Business at

Texas Tech University. She has

published in Communications of

the Association for Information

Systems, Journal of the American

Society for Information Science and Technology, and Journal

of Service Science and Management. Her current research

interests include business value of IT, information system

security, and data analytics.

Jason Triche is the Pat and Judy McDonald Faculty Fellow and

an associate professor of MIS at the

University of Montana. He received his

Ph.D. from the Rawls College of

Business at Texas Tech University.

Triche has ten years of industry

experience in technology and

management consulting as well as

project management and holds a PMP.

His current research interests include

impression management, technology

failures, and data analytics.

Journal of Information Systems Education, Vol. 31(4) Fall 2020

322

APPENDIX A. Final Dictionary (Categories, Subcategories, and Keywords)

1.Technical Skills

1.1 General Analytical

Analysis

Analysis and design

Analysis and reporting

Analysis skills

Analysis support

Analysis techniques

Analysis to support

Analysis tools

Analytic support

Analytical abilities

Analytical methods

Analytical projects

Analytical results

Analytical support

Analytical techniques

Analytical tools

Analyzing information

Quantitative analysis

Quantitative and qualitative

1.2 General Statistics

anova

Advanced statistics

Data modeling

Linear and logistic

Linear models

Predictive models

r-square

Regression

Statistical analyses

Statistical analysis

Statistical data analysis

Statistical methods

Statistical modeling

Statistical models

Statistical process

Statistical reports

Statistical techniques

Statistical tests

Statistics

Summarizing data

1.3 Modeling

Algorithm

Algorithm design

Algorithm development

Algorithms

Algorithms and applications

Algorithms and formulations

Algorithms in solving real

Algorithms to match online

AMPL

Combinatorial optimization

Constraint based

Constraint programming

Cplex

Decision analysis

Decision making

Decision problems

Decision science

Decision sciences

Decision support

Decision support analysis

Decision support applications

Decision support functions

Decision support models

Decision support research

Decision support research analyst

Decision support software

Decision support systems

Decision support tools

Decision tools

Decision trees

Forecasting

Integer

Linear programming

Mathematical modeling

Mathematical models

Mathematical programming

MATLAB

Nonlinear

Non-linear

Optimization

Quadratic

Stochastic optimization

1.4 Model development

Model development

Model formulation

Modeling

Network modeling

Simulate

Simulation

Simulations

1.5 Data Management

Data analysis

Data set

Data collection

Data gathering

Data integrity

Data mining

Database management

Datamart

Dataset

ERD

etl

Large data

Relational databases

Software applications

Journal of Information Systems Education, Vol. 31(4) Fall 2020

323

2. Software Package Skills

2.1 Personal Productivity

Computer skills

Macros

Microsoft Office

Microsoft suite

Microsoft Word

MS Office

MS Word

o365

Power Point

PowerPoint

Spreadsheet

Spreadsheets

Word processing

2.2 Business Intelligence

Cognos

fixml

pentaho

powerbi

tableau

2.3 Database Systems

Access and

And access

cassandra

DB2

dbms

Hbase

Microsoft Access

mongodb

MS Access

Mysql

Nosql

Oracle

SQL Server

*SQL*

Teradata

tsql

*XML*

xsd

xsl

2.4 Stats Packages

SAS

SPSS

STATA

2.5 Programming Language

C++

html

Linux

Pearl

Perl

pig

Python

ruby

VBA

Visual Basic

2.6 Enterprise System

Azure

Google Analytics

Hadoop

Hive

Salesforce

SAP

Watson

3. Academic Degree

3.1 Bachelor’s

Accredited college

Bachelor of business administration

Bachelor of science

Bachelor’s

Bachelor’s degree

College or university

3.2 Master’s

Advanced degree

Doctorate

Graduate degree

Master’s

Master’s degree

MS degree

3.3 Ph.D.

Master of business administration

MBA

MBA degree

PhD

PhD degree

Journal of Information Systems Education, Vol. 31(4) Fall 2020

324

APPENDIX B. Evolution of Software Skills for Data Analysts

Categories and Sub-Categories

2014

2015

2016

2017

2018

Personal Productivity

30.70%

29.70%

32.60%

32.10%

28.60%

Microsoft Office

18.40%

16.30%

20.60%

18.50%

17.20%

MS Power Point

12.10%

13.70%

14.20%

13.90%

MS Excel

4.20%

3.90%

3.70%

5.80%

4.20%

MS Word

3.50%

4.70%

4.30%

3.80%

3.30%

Macros

1.70%

2.40%

1.50%

1.90%

1.50%

o365

0.00%

0.20%

0.10%

0.00%

Database System

48.00%

48.80%

52.40%

58.10%

58.90%***

SQL Server

10.40%

12.10%

13.40%

15.20%

17.80%***

Oracle

15.00%

14.00%

13.40%

14.40%

15.50%

Microsoft Access

13.60%

11.60%

11.30%

11.60%

10.40%**

Teradata

2.30%

2.50%

2.80%

3.10%

MySQL

1.50%

1.80%

3.00%

2.50%

2.10%

XML

2.70%

2.80%

3.00%

2.00%

NoSQL

0.70%

1.00%

1.60%

1.50%

1.60%**

DB2

2.00%

1.90%

1.60%

DBMS

0.30%

0.60%

0.90%

0.90%*

Hbase

0.30%

0.20%

0.30%

0.40%

0.60%

MongoDB

0.20%

0.40%

0.80%

0.40%

0.30%

tsql

0.50%

0.40%

0.60%

0.80%

0.20%

Cassandra

0.30%

0.20%

0.30%

0.20%

xsl

0.00%

0.30%

0.00%

0.10%

0.00%

xsd

0.00%

0.20%

0.00%

Business Intelligence

11.30%

12.60%

17.60%

21.80%

23.20%***

Tableau

5.30%

8.30%

13.20%

18.20%

18.90%***

Cognos

7.50%

5.30%

5.80%

6.00%*

PowerBI

0.10%

1.00%

1.60%

1.90%***

Pentaho

0.40%

0.80%

0.60%

0.30%

Journal of Information Systems Education, Vol. 31(4) Fall 2020

325

JavaScript D3

0.20%

0.00%

0.50%

0.40%

0.20%

fixml

0.00%

0.20%

0.00%

Statistical Package

10.10%

9.50%

12.40%

15.40%

16.10%***

SAS

6.60%

8.00%

8.80%

9.70%***

SPSS

1.60%

1.90%

2.40%

3.70%

2.50%*

4.50%

4.40%

7.00%

9.60%

11.80%***

STATA

0.60%

0.40%

1.10%

0.50%

Programming Language

16.00%

16.40%

20.20%

21.90%

22.50%***

Python

2.30%

3.30%

6.60%

9.10%

10.70%***

9.60%

7.90%

8.60%

9.90%

9.60%

HTML

2.80%

3.00%

3.90%

2.50%

2.70%

VBA

1.90%

2.60%

2.90%

2.60%

2.50%

Linux

1.90%

2.50%

2.00%

1.70%

Visual Basic

1.40%

1.30%

1.40%

1.60%

1.20%

Pig

0.30%

0.50%

0.70%

0.60%

1.00%**

Perl

1.00%

0.90%

0.80%

1.60%

0.90%

Ruby

0.50%

0.30%

0.90%

1.20%

0.30%

C++

0.00%

Pearl

0.00%

0.20%

0.10%

0.00%

Enterprise System

16.00%

15.80%

16.20%

18.50%

20.30%***

SAP

10.40%

9.00%

8.00%

8.40%

8.30%*

Hadoop

2.00%

2.40%

4.00%

4.50%

5.10%***

Salesforce

3.40%

4.50%

2.80%

4.50%

4.90%*

Azure

0.20%

0.00%

0.80%

1.30%

3.60%***

Hive

0.70%

1.10%

1.80%

2.10%***

Google Analytics

0.50%

1.50%

1.00%

1.20%*

Watson

0.40%

0.20%

0.10%

0.30%

0.20%

*p < 0.05, **p < 0.01, ***p < 0.001

Categories are underlined; Subcategories are indented.

Bold (italicized) text indicates a positive (negative) trend between 2014 and 2018.

Journal of Information Systems Education, Vol. 31(4) Fall 2020

326

Information Systems & Computing

Academic Professionals

STATEMENT OF PEER REVIEW INTEGRITY

All papers published in the Journal of Information Systems Education have undergone rigorous peer review. This includes an

initial editor screening and double-blind refereeing by three or more expert referees.

or hard copies of all or part of this journal for personal or classroom use is granted without fee provided that copies are not made

or distributed for profit or commercial use. All copies must bear this notice and full citation. Permission from the Editor is

required to post to servers, redistribute to lists, or utilize in a for-profit or commercial use. Permission requests should be sent to

the Editor-in-Chief, Journal of Information Systems Education, editor@jise.org.

ISSN 2574-3872