empirical. lecture 10. survey research

types of survey

• decriptive

• comparative

• trend survery

• panel survey

two categories of survey types

• cross-sectional

• longtitudinal

cross-sectional

collecting information of responsdents at a single period in time

longtitudinal

surveys that help researchers to collect data over an extended period of time

what types of survey are cross-sectional?

• descriptive

• comparative

what types of survey are longtitudinal?

• trend surveys

• panel surveys

descriptive surveys

for example, asking teenagers' view on cigarette smoking (less theory-driven)

comparative survey

for example, cross-sectional survey about the relationship of peer pressure and cigarette smoking (more theory-driven)

trend survey

tracking different groups of people over time to track overall shifts

panel survey

administered to the same set or group of people over multiple times. somestimes DV is assessed, before IV are assessed (often also for theory testing)

common method bias

error in correlation caused by the measurement method rather than the actual realtionships between constructs

when does common method bias happen

collecting all independent and dependent variable data from the same respondent at the same time.

solutions to common method bias

• panel data, but more difficult to access the same respondent over time. attrition can be high.

• mixing the order of questions/ using different scale types/validated scales.

• DV: measured via archival data (ex: earnings management)

• ex-post: statistical techniques.

• Harman's single factor test

construct validity

refers to the extent to which a survey measures what it says it measures

non-response bias

the bias that results when respondents differ in a meaningful way from non respondents. nonresponse is often a problem witth surveys where the response rate is very low.

factor analysis / dimension reduction

a way to condense a large number of variables into just a few latent variables

things that help with identifying construct validity / internal validity for surveys

• cronbach's alpha (=reliability)

• factor loading (=belong)

• discriminant validity (differentiate)

cronbach's alpha (=reliability)

measure of internal consistency. How closely related a set of items are as a group. Helps to determine whether a collection of iems consistently measure the same characteristic. BEST > 0.50.

factor loading (=belong)

correlation between observed variable and latent factor it’s supposed to measure. loadings tell you which items belong to which factor and how strongly they belong.

Harman's single factor test

statistical check used to detect common method bias.

discriminant validity

tests whether concepts or measurements that are not supposed to be related are actually unrelated.

when does the harman’s single factor test mean that CMB is not a major threat

if the single factor accounts for less than 50% of the total variance.

when does the harman’s single factor test mean that CMB is likely present

if the single factor accounts for 50% of more of the total variance.

threshold for factor loading

items under 0.4 should be dropped

what does a high Cronbach’s alpha indicate?

that response values for each participant across a set of questions are consistent. if participants give a high response for one of the items, they are also likely to provide high responses for the other items.

what does a low cronbach’s alpha indicate?

the set of items do not reliably measure the same construct.

thresholds cronbach’s alpha

• o = no correlation between items.

• 0.5 = reasonably reliable

• 1 = perfectly correlated. questions might be redundant.