A scale critique of impression management by Bolino and TURNLEY (1999)

Study 1: pilot study to content validate
Study 2: improved scale given to Professionals and then exploratory factor analysis completed
Study 3: improved scale given to Managers, then new exploratory factor analysis completed
Study 4: final version of the scale given to a new sample of professionals, but this time confirmatory factor analysis was performed.
Study 5: Using mgt students, the convergent and Discriminant validity scales were assessed.

Issues with previous research on impression mgt (I.M) by Rao, Schmidt, and Murray (1995)
1) research focuses on just a few I.M strategies
2) an overuse of lab studies using student samples
3) there is a shortage of empirical assessments of developed I.M framework

The need for this scale, according to the authors was that "no widely accepted measure of impression management behaviors" existed.

Previous approach to measuring I.M=
1) "observing and recording participants' impression management behaviors in an experimental context or under naturally occurring conditions."

The issue is getting into orgs to watch Behavior of I.M

2) "the use of one of the impression management scales developed by Wayne and Ferris (1990) or Kumar and Beyerlein (1991)"

Historical I.M scales I. Wayne & Ferris - 24 item scale, monitors number of times an individual enters into "supervisor focused, self-focused, and job focused impression management behaviors".
Scale strengths:
It relies on self-report data on I.M rather than raters perception of an individuals I.M. It is also easy to administer, which is useful for organizations.
Scale weaknesses:
1. Psychometric problems with one of the impression management Sub-scales, the self-focused sub scale f I.M. It has poor reliability and many items do not show strong Discriminant validity (Ferris et al. 1994)
2. The I.M scale was concluded from just exploratory analysis, so the scales exact meaning is unjustified.
3. I.M Subscales "intimidation" and "supplication" are missing
4. An overlap of items measuring two different constructs

2nd measure of I.M
Kumar and Beyerlein's (1991) Measure of Ingratiatory Behaviors in organizational settings (MIBOS)
Weaknesses:
1)Focuses on just one specific form of I.M (ingratiation)
2)Kacmar and Valle (1997) criticized the validity of the MIBOS
3) overlap with items measuring a conceptually different construct (OCB: org citizenship behavior)
4) they fail to deal with the issue of motivational intent, and can therefore not distinguish between impression management behavior and OCB. Meaning that if there is no motivational intent, then it may just be OCB at work. If there is motivational intent then the worker may be using impression management tactics.

Research has really only studied 2 of the 5 scale areas suggested by jones and Pittman: ingratiation (like ability), and self promotion.

This study was created to measure all 5.

The authors used Hinkin's (1998) 6 step framework to develop their scale over 5 studies item generation, (b) questionnaire administration, (c) initial item reduction, (d) confirmatory factor analysis, (e) convergent/discriminant validity, and (f) replication.
Isn't there a rule that we only really accept eigenvalues over 1.00? If that is the case then this is a 3 factor model.

That said sometimes that rule is wrong right?
Inter-factor correlations show independence of factors 1 against 2,3 & 5
2 against 1 & 4
3 against 1
4 against 5

So there is some overlap between factors
1 and 4
2 and 3 and 5
3 and 2 and 4 and 5
4 and 1 & 3

Confirmatory Factor Analysis

Does each individual item load heavily on its own corresponding latest factor? The items, the loadings, and the inter-factor correlations were displayed.

Inter-factor correlations were run across the 5 versions of the I.M scales.
High correlations were found between factors "intimidation" & "supplication";
"Ingratiation" & "Exemplification";

The estimated fit of the 5 factor model of I.M was run against alternative solutions. It was the aim of the authors to avoid finding an alternative model fitting better than their 5 factor model. They used CFA to test alternative factor structures, especially for those factors that had inter-factor correlations above r< .50

They ran two 4 factor models instead, each one combining one of the pairs noted above into one factor (I.e intimidation" & "supplication).
They created a 3 factor model with both high correlated pairs merged into 2 factors and leaving "self-promotion" as the third factor.

Finally they tested to see if Impression Management was in fact a multi-dimensional scale, but running a 1 factor model to see if it fits well.

Results showed that in all cases, a five factor model was the best fit when compared to all other alternative models

Extra validity test
Higher-order Confirmatory Factor Analysis was run on the five factors latent variable level, to test for a global impression management factor.
"The fit indices for the higher-order model were good (GFI = .91, TLI = .92, CFI = .94)", and all sub-factors had paths that linked to the global factor (the mother factor, think hierarchical factor) global impression management factor paths were .48, .62, .78, .46, and .65

Reliability measures:
Cronbach's alpha was used for each factor, and based the minimum alpha level on Nunally's (1978) suggestion of .70 or higher to be acceptable alpha.

The authors found the following alphas: self-promotion (alpha = .78), ingratiation (alpha = .83), exemplification (alpha = .75), intimidation (alpha = .86), and supplication (alpha = .88).

Convergent & Discriminant validity
In study 5, using working students, the final I.M scale was given to them along with measures of: organizational citizenship behavior (OCB), conscientiousness, perceived organizational support, self-monitoring, and careerism.

OCB was assessed using Van Dyne, Graham, and Dienesch's (1994) scale.
Conscientiousness was measured with Goldberg's (1992) 20-item scale
Perceived organizational support was measured with the shortened version of the Survey of Perceived Organizational Support (Eisenberger, Huntington, Hutchinson, & Sowa, 1986).
Self-monitoring was measured using Lennox and Wolfe's (1984) 13-item scale.
Careerism was measured using eight items taken from Feldman and Weitz (1991)

Support for convergent validity:
Self monitoring and careerism significantly related to I.M dimensions, and not to OCB.
"Specifically, self-monitoring was positively and significantly correlated with self-promotion, ingratiation, and exemplification. Self-monitoring was not significantly related to any of the dimensions of OCB. In addition, careerism was positively and significantly correlated with self-promotion, intimidation, and supplication. Also, careerism was negatively related to the loyalty dimension of OCB."

Support for Discriminant Validity:
"conscientiousness and perceived organizational support (traditional predictors of OCB) were not significantly related to any of the specific impression management tactics. Although, again as expected, both were generally related to the specific dimensions of OCB."
In addition, the Subscale of the I.M measures don't match the Subscales of the OCB measures. Out of the 15 inter-factor correlations of both measures dimensions, only 3 were found to be significant.
"Specifically, obedience was negatively related to supplication, and functional participation was positively related to ingratiation and exemplification"

Additional strength of the instrument was being developed using diversified samples from a range of organizations (total N = 698)

Limitations of the scale noted by authors:
1) more research needs to be collected to verify that all Hinkin's (1998) scale development guideline processes have been met.
2) more replications are needed of this scale, to create norms for the measure
3) test-retest measures of reliability should be researched to see the impact of this scale over time on the individual
4. The authors believe there could be a social desirability bias threat in their data, because the scale items include motivational intent. So they suggest future studies look at if social desirability bias can impact individuals reporting impression management behaviors.

Ideas for future research:
1) look at the antecedents of the 5 strategies of I.M and look for any patterns by which they exist.
2) study if and how each I.M strategy is used to achieve a certain goal of eliciting an attribution from others
3) alternatively study if and how using I.M strategies to influence others can backfire
4) more research is needed to compare I.M to organization citizenship behaviors
5) evaluating the impact of OCB and I.M on performance ratings and promotions

KACMAR, K. M., HARRIS, K. J., & NAGY, B. G. (2007). FURTHER VALIDATION OF THE BOLINO AND TURNLEY IMPRESSION MANAGEMENT SCALE.

IM defined by Rosenfeld, Giacalone & Riordan (1995) "IM is the process by which individuals present information about themselves to appear as they with others to see them."

The reason Impression Management scale was developed by Bolino was the lack of validity for the 4 dimensions out of the 5 they suggest. Only "ingratiation" has been well researched (see meta-analysis by Gordan, 1996), based on the well validated ingratiation scale, developed by Kumar & Beyerlein, 1991.

One issue with the Bolino I.M scale is that the final sample to test the final version of the scale, consisted of a student sample, that may not be reflective of the working world. Thus all 3 samples used in Kacmar's study were of full-time workers (1 sample of supervisors, 1 sample of subordinates, and 1 sample of a mixture of subordinates and supervisors)

Kacmar adds "self-promotion" to the factor combo from Bolino's study: ingratiation/exemplification as it also posits the "actor" in a positive light.

To test convergent validity, Kacmar uses an I.M scale called "IMAS" (Andrews & Kacmar, 2001) to compare with Bolino scale, which consists of 4 factors (Blaring, Blurring, Boasting, & Burying). Actors either advertise positive (Blurring or Boasting) or they hide the negative (Blaring & Burying). It was their hope that the two measures would correlate.

To test discriminant validity, Kacmar compares Bolino scale to OCB scales, but uses two different sub-scales of OCB: Altruism & Citizenship.

Criterion-related validity used included the subjects perception of politics in the organization as a criterion variable. The idea being that a subordinate may respond to "a perceived political environment" in a way that uses ingratiation to obtain higher ratings from supervisors, or attempt "supplication" to gain some added support from others in their team. Their hypothesis 5, suggested that the 5 "I.M dimensions would positively predict perceptions of organizational politics".

Issues with Kacmar's study:
They used all full-time workers whilst trying to replicate and test Bolino's study, which means it is not really a replication of Bolino's scale.
Although the recipients are told the data will be confidential and not reported to their managers, it is not anonymous. Names and numbers are kept to call subjects to confirm them are completing the study. There is a danger of this having an effect on the results, and their feeling of how secure their response will be.
They did not use the same sample as Bolino, which is really unavoidable, but still an issue in replicating the scale.

Reliability of Measures:
Kacmar ran internal consistency reliability estimates (cronbach's alpha) across the three samples they used, and reported strong reliability across the 5 dimensions of I.M (shown below)
Self promotion: (alpha = .88, .86, and .92)
Ingratiation: (alpha = .91, .85, .91)
Exemplification: (alpha = .81, .79, and .76)
Intimidation: (alpha = .87, .89, and .84)
Supplication: (alpha = .93, .93, and .93)

Then to show reliability for the IMAS, the same exercise was conducted.
Blaring (alpha = .82)
Blurring (alpha = .81)
Boasting (alpha = .83)
Burying (alpha = .81)

To measure the political perceptions, Kacmar used the Hochwarter, Kacmar, Perrewe and Johnson (2003) measure, with 6 items.
Political Perception (alpha = .94)

Finally the two extra OCB measures were calculated for internal consistency.
Altruism, a 3 item measure (alpha = .86)
Citizenship, a 3 item measure (alpha = .87) as opposed to Bolino's OCB measures (loyalty, obedience, and functional participation)

They then ran 7 fit models, 4 of which were their own alternative combinations of the scale dimensions suggested by Bolino. These alternative models included 2 types of 4-factor models, and 1 type of a 2-factor model. When compared to the Bolino's 5 factor model, and 1 factor model, the 5 factor model showed fit statistics (CFI = .95, NFI - .91, RMSEA = .08) which were all significant and supported the idea of a 5 factor model being the best fit.

Then running Lisrel, a confirmatory factor analysis was run to show the 5 factor model holding across the three samples. Results indicated two items with a low loading (.51) from the "exemplification" factor, and (.61) from the "Intimidation" factor. They decided to only focus on the .51 as the weak link that was also found in Bolino's factor analysis across each of their samples. Later Kacmar suggests replacing the item, or removing it completely, but makes no mention of the .61 item, even though using Hinkin's descriptive, Anything under .7 is probably no good to us as an item. Below .7 it indicates high variance amongst responses, making our construct measurement less reliable. Kacmar also suggests that by correcting problems with this item, the overall reliability should rise on the "exemplification" scale.
Kacmar suggests adding a name to the scale, to add purpose to the scale, and aid future research efforts.
Suggestions include: IM-5, BTIM-5 (after the initials of authors), or SISIE which is an acronym of combining all 5 dimensions first letters.

Results indicate every I.M scale item from Bolino and every IMAS item correlated significantly, except the relationship of "ingratiation" and "Blurring" (r=.01). "Boasting" has the highest correlation with the I.M scale (.44)(.30)(.40)(.32)(.34) although these numbers are generally a lot lower than was hoped for by the authors, therefore this is an issue with the convergent validity.

Discriminant validity was supported against all 5 dimensions of Bolino's I.M scale
Altruism (.04)(.11)(.10)(.07)(.03)
Citizenship (.04)(.06)(.04)(.04)(.01)

Support for the criterion-related validity (perception of politics) is shown by running a hierarchical regression controlling for variance due to gender and age (both have theoretical relationship with perception of politics). Results showed that 4 of the 5 I.M dimensions did explain "significant incremental variance in the perception of politics" with "self-promotion" not being significantly linked.

Future variables are suggested to be studied to add convergent and discriminant validity, including:
1. social desirability;
2. antecedents: need for power, self monitoring, self esteem,
3. logical outcomes: performance and promotability

Other suggestions for future research:
More criterion-related validity and incremental validity studies should be run.
Use of the full I.M scale should be used when "exploring employee reactions to the organizational workplace"
For researchers studying org justice, it would be interesting to see if I.M will increase in usage depending on how the worker perceives injustice at their organization
Can co-workers and supervisors actions influence the work environment to the degree that I.M tactics are necessary, encouraged, or just not spoken about?

References:
Kacmar, K. M., Harris, J. K., Nagy, G. B. (2007). Further Validation of the Bolino and Turnley Impression Management Scale. Institute of Behavioral and Applied Management.

Laura Morgan Roberts, (2006), Changing Faces: Professional Image Construction in Diverse Organizational Settings, Academy of Management Review.

Bolino, M. & Turnley, W. 1999. Measuring impression management in organizations: A scale development based on the Jones and Pittman Taxonomy. Organizational Research Methods, 2: 187-206