MGMT 442 - HRIS and People analytics ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Attend anywhere Participation grade Reading / video before almost every class Weekly announcements posted every week Weekly open note quiz Office hours 12- 1 pm tuesday + thursday Have assign readings Extra credit opportunity participating in university research Late assignment policy 24 grace period with data exercises - email night of due each day late after is ten percent reduction Four things due by sunday Data exercises take 30 min to an hour usually HRIS ○ A system used to acquire store analyze and or distribute pertinent information about an organizations human resources Goals of hris ○ Provide accurate and timely data Hris importance ○ Comprehensive info into integrated databases ○ Improve hr operations + mgmt operations ○ Facilitate user decision making ○ Shift focus from transaction to transformative action ( atomization plays a big part ) Hr analytics ○ Process of collecting analyzing and reporting people related dat for the purpose of improved decision making and achieving strategic direction Hr analytics project life cycle ○ Question formulation - data acquisition - data management - dat analysis - data interpretation and storytelling - deployment and implementation Evolution of hris and hr analytics Limits of tech ○ Acquiring and using low quality data ( bias, stale, inaccurate) data will lead to problems Need for ethical decision making Competencies needed for hr analytics ○ Hr expertise / data literacy 1/6/2022 ● ● Tutorials + help in canva Tip watch material before taking checki-in quiz ● Four assignments due by sunday at 11:59 pm ● Notes ● Question formulation : process of posing strategy inspired research questions ● Good questions = better data ● Hypothesis - statement about expected nature of a phenomenon of interest ● Install R - watch canvas chapter 9 This weeks links 1/11/2022 ● ● ● ● ● ● ● ● ● ● ● ● Data exercise 2 and week 2 knowledge check in due by sunday Read before class from now on and take notes Data acquisition - collecting, retrieving, gathering and sourcing info ○ Done through ■ Employee surveys ■ Rating forms ■ Surveillance monitoring ■ Database queries ■ Scraping Employee surveyed - good for observing perceptions, attitudes, job satisfaction, ○ Disadvantages - more subjective in nature than some other tool, influenced by social desirability Rating forms - similar to employee surveys, tend to be more focused on measuring performance ○ Disadvantages - achieving high reliability can be challenging across raters Surveillance and monitoring ○ Nonintrusive and operate behind the scenes ○ Tech improving rapidly with capabilities to measure geolocation, tone of voice, interactions, heart rate, sleep quantity/quality etc. ■ Disadvantages ● Invasive, privacy rights, Database queries - way to gather already present data and manage and leverage it ○ Disadvantages - quality of data, lack there of data, important characteristics unless clearly defined may be hard to find Scraping - new tools make it easier than ever, new insights into previously difficult-to-reach data ( could do it with twitter data - r has packages to do it) ○ Accuracy, invasion of privacy - erosion of privacy Guiding principle with data acquisition acquire data with a purpose HRIS Analysis Systems development life cycle ○ Planning - analysis - design - implementation - maintenance Needs analysis - gathering, prioritizing, and documenting and orgs requirements ● ● ● ● ● ● ● ● ● ● ● Five stages of need analysis ○ Planning - looking at current system and what we need ○ Observation ○ Exploration - build on analysis from observation stage, gather additional data ○ Evaluation - review and asses data collected ○ Reporting - prepare and report summarized findings. 1/13/2022 Design Two step process logical ->physical design Logical design - translate business requirements into improved processes, irrespective of tech implementation ○ For example applicant racking process Process models ○ Include things like data flow diagram ■ Which is a graphical representation of key business activities and processes in the hr system ■ Context level diagram - describes full system, boundaries, and the entities the system interacts with ■ Level 0 diagram - greater level of detail then context level ■ Additional level diagram Data flow diagram symbols ○ Rectangle represents an external agent or entity ○ Arrow represents data flow ○ Rounded rectangles represent process or business activity ○ Open ended rectangle represents data store, data at rest For charting ○ https://www.lucidchart.com/pages/ Physical design ○ Determine the most effective means of translating the business process into a physical system that includes hardware and software Physical design three choices ○ Do nothing ○ Make changes to only the hr business processes without new tech ○ Implement process change with new or upgraded tech Benefits and drawbacks of building hr system internally versus outsourcing ○ You can make it specified to your needs ○ More flexible in terms of changing ○ The ability to integrate when making your own system ○ Huge upfront cost ○ Need access to skills and ability to do it Working with vendors ○ Org develops request for proposal ○ Venderos review rfp and provide feedback ● ○ Org evaluates vendors reviews ○ Org selects vendor Assessing system feasibility ○ Tech feasibility ○ Operational feasibility ○ Legal and political feasibility ○ Economic feasibility 1/18/2022 READ BEFORE CLASS!!!! ● ● ● ● ● Change management ○ Systematic process of applying knowledge, tools, and resources to transform organizations from one state of affairs to another Addressing resistance to change ○ Perceived need ○ Upper management buy-in ○ Timely change conversion ○ Clearly communication of change ○ Provide proper training Assessing Individual change ○ There are different theoretical models that can help assess preparedness for change ■ Readiness for change model ( transtheoretical model ) ● Precontemplation \/ ● Contemplation \/ ● Preparation \/ ● Action \/ ● Maintenance ■ Intention to enact behavior ( theory of planned behavior ) ● attitude , norms, control \/ ● Intention \/ ● Use HRIS Thursday we have a case study for the first part of class Logical operators ○ < less than ○ > greater than ○ <= less than or equal to ○ >= greater than or equal to ○ == equal to ○ != not equal to ○ | or ○ ○ & and ! not 1/27/2022 ● Rewatch lecture from tuesday + take notes ● Examples of HRIS applications Does our company meet industry averages in terms of equity and inclusions for such protected classes as disability status, race, sexual orientation etc? 2/3/2022 ● Watch lecture from tuesday 2/1/2022 ● Measurement Scales Practice ○ Length ( ordinal ) ■ O-1 years ■ 2-4 years ■ 5-10 years ■ 10+ Types of Descriptive Statistics Categorical - variables can be described using counts/frequencies Continuous ○ Indices of central tendency ■ Mean ■ Median ■ Mode ○ Indices of dispersion/variability ■ Range ( highest score minus lowest score ) ■ Variance ( collective distance of value difference ) ■ Standard deviation Interquartile range ○ Lower quartile ¼ middle 50% upper quartile 75% ● ● ● ● 2/10/2022 ● ● Writing Survey Items Things to Do ○ When writing employee survey items, do the following : ■ Use clear, simple language ■ Keep the survey as short as possible ■ Consider the expectations you might create with your items ● ● ● ● ● ● ● ● ● ● Things to Avoid ○ Too cute or too clever ○ Unanswerable by employees ○ Too heavily reliant upon open-ended questions ○ Double-barreled ○ Loaded or leading ○ Non-specific or too broad ○ Negatively worded ○ Yea saying or nay saying Things to consider ○ Put careful thought into writing items that are; ■ Required or force a response Developing a measure for a construct ○ Measure are used to gather observations (data) related to a construct ○ But we never perfectly measure a construct ( or a person’s true level of something ) ○ In general, it is a best practice to use multiple items to measure a single psychological construct - and to avoid single - item measures when possible Developing a Measure for a construct ○ Steps to follow for developing construct measures ■ Carefully define the construct ■ Discuss the meaning of the definition and construct ■ Write many items that are in terms of content and redundancy ■ Pilot a large number of items than you plan to eventually use ■ Use a factor analysis to reduce the number of items ■ Determine the internal consistency reliability Reliability ○ Extent to which a measure consistently or dependably measures something Validity - accuracy Conventionally, a measure demonstrates an acceptable level of reliability when a reliability estimate is .7 or higher, where .00 indicates very low reliability and 1.00 indicates very high reliability Strive for reliability to be as close as one Different types of reliability ○ Test-retest reliability ○ Inter-rater reliability ○ Internal consistence reliability Internal Consistency Reliability ○ A reliability estimate based on the intercorrelation ( homogeneity) among items on a test. With alpha being a prime example. ■ For example, if a measure consists of four items/questions and is administered to a group of applicants cronbachs alpha can be computed for the scores on those four items and serve as an estimate of reliability ○ Cronbachs alpha ■ ■ ■ ■ ■ ■ ● ● ● ● ● ● .95-1.0 is excellent .9-.94 is great .8-.89 good .7-.79 acceptable .6-.69 questionable .0-.59 unacceptable Survey administration ○ Survey administration goals ■ Pilot survey - small but representative group ■ Obtain a representative sample of respondents from the employee population - ( how to reduce sampling error ) ■ Maintain employee trust ■ Analyze data in timely manner to avoid “stale” data ■ Be ready to act on survey findings even if the findings look unfavorable to the organization ■ Train managers on what to do with the findings based on their team result ■ Consider how frequently to survey - once a year vs “continuous listening” Survey administration ○ Respondent faking and social desirability ○ Anonymity vs confidentiality vs identified ○ Continuum of creepiness Faking and social desirability ○ When responding to survey items, respondents might distort or fake their responses to conform to what they think is socially desirable or to what they think decision makers would like to see ■ How to address it ● Tell them to answer honestly ● Tell them faking can be detected ● Use lie-detection items ( “ I always practice what I preach” ) Anonymous vs confidential vs identified ○ Anonymous ○ Confidential ○ Identified A continuum of creepiness? ○ How creepy is the going to be for our employees? Advanced Survey Techniques ○ Surveys aren't the only way to gather information ■ Focus groups & interviews ( ie., exit interviews ) ■ Look to the literature ( Research ) ○ Employee sensing ( data mining, ie emails ) ■ Obvious ethical concerns ■ Requires more advanced analyses like natural language processing ○ ○ Use AI and machine learning ■ To automate analysis, reporting, and individualized recommendations Emotive analytics ■ Measuring attitudes via touch, image, or speech 2/15/2022 ● ● ● ● ● ● ● ● ● ● ● ● ● ● Experimental design ○ A study in which an intervention is deliberately introduced to observes its effects Effect ○ Observed difference between what did happen when employees were exposed to intervention and what would happen if those employees simultaneously had not been exposed to the intervention Our goal with experimental design is to approximate a counterfactual using experimental design True experiment design ○ Employees are randomly assigned to a treatment or control condition, with key characteristics being random assignment and control group Quasi-Experiment ○ similar to true experiment but lacks random assignment to conditions Pre-experiment ○ Lacks random assignment and control group - but includes an intervention Nonexperiment design (observational, correlational) ○ Lacks random assignment, control group, and deliberate intervention. Typically just involves observing association between variables Confidence in casual inference decreases from true experiment to nonexperiment design Experimental design is a type of research design Common training evaluation research designs ○ Post-test only without control group ○ Pre-test, post test without control group ○ Post test only with control group ○ Pre-test,post test with control group Statistical and practical significance ○ Statistical significance ■ Addresses whether a hypothesized association or difference exists in underlying population Practical significance ■ Addressees the size of an association or a difference Null hypothesis significance testing Inferential statistics ○ Are valuable ■ Because results are based on a single sample size ■ Date are not often based on entire population scores ■ There is a measurement error and sampling error Inferences about population ● ● ● ● ● ○ Must infer whether the statistics accurately represents the population Two types of hypothesis ○ Research Hypothesis ■ association between the two variables ○ Null hypothesis ■ No association between the two variables Statistical significance: p-value ○ P-value used to reject or fail to reject the null hypothesis ■ Example - when testing null hypothesis that two means are equally and we find a p-value of.03 ■ Typically alpha level represents the cutoff point for statistical significance ● Conventionally alpha is .05 or 5% probability ● For our purpose we’ll use a two tailed test ○ False positive ( ■ We reject the null hypothesis based on the p-value, but the null hypothesis is actually true ○ False negative ■ We fail to reject the null hypothesis, but the null hypothesis is actually false Confidence intervals ○ Provides a range of values that likely contains an unknown population parameter; represents the margin of error around a point estimate. Practical significance - only focus on if significant ○ Effect size can be used as indicators of practical significance ■ Examples of effect size ● Cohen’s d ● Cohen’s f ● r ● R2 ● Odds ratios Watch rest of video for r-studio 2/17/2022 ● ● Criteria for causation ○ For x to be interpreted as a cause of y ■ Y must not occur until after x ( temporal precedence ) ■ X and y are associated with one another ( cooccur ) ■ Other explanations of the association between x and y can be eliminated ○ Rewatch r for todays lecture For this class, when the p-value is less than .05, we reject the null hypothesis and thus conclude that an effect or relationship is statistically significant. 2/22/2022 ● Review of correlation ● ● ● ● ● ● ● ● ● ● ● ● Correlation ○ When wish to estimate a pearson product-moment correlation, both variables should have an interval or ratio measurement scale A correlation coefficient can range from -1 to 1 A correlation coefficient equal to zero equals no linear association between two variables If the p-value associated with a correlation coefficient is less that .05 then the association is significantly significant Importance of validity in selection High reliability is a necessary but not sufficient condition for high validity Validity ○ Is the degree to which available evidence supports inferences made from scores on selection measures Validation study ○ Research process used to investigate how well a selection tool measures a particular construct and whether scores on the selection tool are job related ■ Types ● Criterion related validity ○ Concurrent ○ Predictive ● Content validity ● Construct validity ○ Convergent ○ Divergent Criterion-related validity ○ Degree to which scores are related to scores on a criterion ■ Criterion could include ● Manager ratings of job performance ● Customer satisfaction ratings ● Absenteeism ● Theft ● Accidents In selection we can use correlation and regression to assess the criterion-related validity of a selection Concurrent validation study ○ Selection tool given to people on the job and correlate it with the same criterion at the same time Predictive validation study ○ New applicants given the selecting tool, but dont use to hire, and several months later, correlate it with the criterion scores of those who were hired ● 2/24/2022 ● ● ● ● ● ● ● ● ● Simple linear regression models include one predictor variable and one outcome variable The regression coefficient associated with a predictor variable represents the linear association between the predictor and outcome variables The model intercept value represents the value on the outcome variable when the predictor variable is equal to zero The r2 value represents the model’s fit to the data or the proportion of variance explained by the predictor variable in the outcome variable Yhat = b0 +b1x1 ○ B0 is the model intercept ○ Yhat is the predicted score on the outcome variable ○ B1 represents the regression coefficient ○ X1 represents the predictor variable Predict-ish analytics ○ Predict what will happen in the future based off a snap shot from the pass without verification ○ Short coming - tendency to overfit model, making it unlikely to work well with new (future) data Predict-ish analytics process ○ Past data ■ Data can b e acquired from a concurrent validation study ○ Build model ■ We can estimate a simple or multiple linear regression model ○ Results ■ The regression model coefficients can be used to construct a predictive equation, for which applicant scores on the selection tool could be plugged in Predictive analytics ○ Predict-ish analytics with verification Predictive analytics process ○ Phase one ■ Past data ■ Build model ■ Results ○ Phase 2 ■ New data ■ Apply model ■ Predictions 3/1/2022 ● ● Can remove one task from team project Multiple linear regression ○ ● ● ● ● ● ● ● Regression coefficient is the linear association between the predictor and outcome variables when statistically control for the effects of other predictor variables ○ R2 value represents the effect size or the collective variance explained by the predictor variables Multiple linear regression equation ○ Yhat = b0 + b1x1 + b2x2 ○ B0 = y-intercept ○ Yhat = predicted outcome value ○ B1 b2 represent the regression coefficients predictor variables ○ Watch video finish this part Incremental validity ○ Evidence of incremental validity can signify that a selection tool explains unique variance in the criterion when accounting for the effects of other selection tools Multi tool selection systems ○ Any instance when more that one selection tool is used to select candidates Different multi tool selection systems ○ Multiple hurdle approach ○ Single hurdle ■ Compensatory ■ Non compensatory Multiple hurdle - multiple stages where people have to make it through different stages Compensatory approach ○ Adds weighted scores to different predictor variables with the application of multiple linear regression ( with the regression coefficients ) Noncompensatory ○ Bench mark scores need to be met to move through the hurdles ( not able to compensate for deficiencies from selection tool to selection tool ) ● 3/3/2022 ● ● ● ● ● Model & algorithms Model is a parsimonious representation of phenomenon An algorithm can be described as a set of rules often laid out in some process We must keep in mind that a model is inherently subjective as somewhere during the decision humans made decisions regarding ○ What datas used ○ What features are included in the model ○ What algorithm is used ○ How to interpret the output of the model Ai and selection tools ○ Increasingly, vendors like Hirevue are offering ai-based assessment products that can be used for recruitment/selection ● ● In 2022 the society for industrial and organizational psych published this set of guidelines for evaluating ai-based assessment tools Compensatory approach to selection decisions ○ Uses applicant selection scores from multiple tools - helps compensate for weak areas potentially 3/08/2022 ● ● ● ● ● ● ● ● ● ● ● Data Privacy and Security Ethics ○ Standards and practices that tell us how human beings ought to act in many situations in which they find themselves ○ Frame work for ethical decisions ■ Id ethical issue ■ Gather facts ■ Consider and eval alternatives ■ Select decision alternative Data privacy ○ Individuals ( perceived ) control over the collection, storage, access, and reporting of their personal data According to a study by eddy et all people perceived the use of hris as invasive when ○ Supervisors are able to access the information in the employee records ○ Employees do not have the ability to check the accuracy before the decisions were made Scaping data ○ Technique used to extract data from websites and other text documents ( what i could do for crypto ) ○ Can be rich info but is also a point of contention Data Security ○ As of 2015 some sources claim that over half billion sensitive organizational records have been breached since 2005 3/10/2022 Storytelling with data Strategic storytelling ○ “ in my work at nasa i coach leaders on how to tell stories that accelerate positive change.” Understand the context to narrow audience ○ Who ○ What ○ How Choose an appropriate visual display Data visualization ○ Is a set of processes via which data are graphically displayed and interpreted with a particular goal in mind to ultimately derive meaning ● ● ● ● ● Minimize cognitive load Focus attention where you want it Attention focusing consideration ○ Size denotes relative importance ○ Use color sparingly ○ Use color consistently ○ Color evokes emotion ○ Zigzag of taking in info Think like a designer ○ Four A’s of design ■ Affordances ● Aspects inherent to design that make it obvious how a product is to be used ■ Accessibility ● Designs should be usable by people of diverse abilities ■ Aesthetics - make it pretty ● People perceive more aesthetic designs as easier to use than less aesthetic designs ■ Acceptance - make sure people like it ● For a design to be effective, it must be accepted by intended audience Good storytelling ○ Find subject you care about ○ Don't ramble about subject ○ Keep it simple ○ Have guts to cut it ○ Keep your tone ○ Say what you mean to say ○ Pity the audience