Bataineh and colleagues measure perceptions of university websites
All variables in the 791-response Jordan study are self-reported; its ranking construct measures perceptions and preferences (Bataineh et al., 2025, §3.2, Appendix A).
The 791-response study measures people’s perceptions of university websites, including what it calls search engine rankings. It does not independently record Google positions for the sampled institutions. Bataineh, Qasim, Idris and Abu-AlSondos, 2025, §§3.2 to 3.3 and Appendix A.
That outcome definition deserves to come before the path coefficients. The ranking questions ask about visiting first-page results, associating high rank with credibility and trusting highly ranked sites. Those are meaningful attitudes. They are not a server log, a search-results capture or a demonstrated change in rank after a website edit.
Who responded, and how were they recruited?
The authors selected five private universities across geographic strata in Jordan: Jadara University, Zarqa University, Applied Science Private University, Al-Ahliyya Amman University and Al-Zaytoonah University of Jordan. Within the selected institutions, they used convenience sampling because of access and resource limitations. It would therefore be inaccurate to describe the respondents as a simple random sample of every university user in Jordan. [§3.1, PDF p. 6, printed p. 5].
Collection ran from February to April 2024, using a structured questionnaire distributed through university social-media links and campus QR codes. Table 1 reports 791 participants. The earlier questionnaire check involved 30 participants; it is a separate instrument-development step rather than another independent outcome sample to add to the reported base. [§3.3, PDF p. 7, printed p. 6].
| Role in the reported sample | Respondents |
|---|---|
| Student | 508 |
| Faculty member | 202 |
| Administrative staff | 81 |
| Total | 791 |
Source: Table 1, PDF p. 7. These are sample counts. Recruitment through available links and campus access limits population inference; a large sample does not remove that selection issue.
There is a reporting inconsistency later in the paper. Its limitations section describes a sample confined to students and proposes adding faculty and staff in future work, even though Table 1 includes both groups. We retain the actual role counts and identify the conflict. It does not justify guessing whether those groups were excluded from some undocumented analysis. [§5.2, PDF p. 14, printed p. 13].
What did the questionnaire measure?
The six constructs are content quality, technical SEO, backlink quality, mobile friendliness, user engagement and search engine rankings. Each uses five agreement items on a five-point Likert scale, with English and Arabic versions. These are instrument design choices, not observed proportions to which a Wilson interval would apply. [§3.2].
Appendix A makes the meaning of the constructs concrete. Content questions concern relevance, originality and informativeness. Technical questions include perceived loading speed and navigation. Backlink questions ask about credibility and the respondent’s willingness to visit a frequently referenced site. The engagement items concern interaction, time spent, return visits and recommendations.
The ranking construct is especially easy to mislabel. It combines attitudes toward first-page results, low-ranked sites, credibility, trust and authority. A model explaining that construct may be explaining related perceptions. It is not necessarily explaining variation in a search engine’s actual ordered results.
The authors acknowledge “self-reported measures for all variables.” They also discuss the possibility that collecting predictors and outcomes through the same questionnaire can influence their relationships. That is a relevant limitation even when reliability and discriminant-validity checks meet the authors’ thresholds. [Bataineh et al., §5.2].
What do the model’s reported paths say?
| Table 7 path | Reported coefficient | Reported t |
|---|---|---|
| Content quality to engagement | 0.231 | 5.062 |
| Technical SEO to engagement | 0.196 | 4.037 |
| Backlink quality to engagement | 0.186 | 4.013 |
| Mobile friendliness to engagement | 0.189 | 4.334 |
| Engagement to ranking construct | 0.623 | 26.774 |
| Content quality through engagement to ranking construct | 0.144 | 4.975 |
| Technical SEO through engagement to ranking construct | 0.122 | 3.937 |
| Backlink quality through engagement to ranking construct | 0.116 | 3.901 |
| Mobile friendliness through engagement to ranking construct | 0.118 | 4.292 |
Source: Bataineh et al., Table 7, PDF p. 11, printed p. 10, reported analysis of the 791-response sample. The table prints p values as 0.000. That is reporting precision, not a literal probability of zero. These are fitted model coefficients and test statistics, not counts or percentages of websites; Wilson intervals are not applicable.
The direction of an arrow is part of the hypothesized model. Fitting a positive path does not by itself establish that changing its starting variable causes a change in the ending variable. A respondent who generally likes a website may score both engagement and credibility highly. Reverse relationships and omitted influences remain possible.
This matters for the strongest-looking path. The engagement-to-ranking coefficient is not a measured gain in Google position from increasing time on site. Both sides of the relationship are self-reported constructs. Converting the coefficient into a ranking-factor weight would change the measurement and the design simultaneously.
Which printed numbers conflict?
For the indirect technical-SEO relationship, Table 7 reports 0.122 with t = 3.937. The preceding narrative instead prints 0.231 and t = 5.062 for that mediated relationship, matching the direct content-quality values elsewhere. We use the table value when describing the reported model and keep the conflicting prose visible. [§4.2, PDF p. 10, printed p. 9; Table 7, PDF p. 11].
The cause is unestablished. We do not claim to have corrected the underlying model by selecting the more plausible number. A definitive reconstruction would require the response-level data, the measurement specification and the fitted analysis outputs.
Table 6 also gives R² = 0.388 for the ranking construct and 0.435 for engagement. These describe explained variation in the fitted latent-variable models, conditional on their definitions and sample. They are not the proportion of Google ranking behavior caused by the included practices, nor an independently tested out-of-sample forecasting result. [Table 6, PDF p. 10].
The paper’s prose sometimes describes the model in stronger language about influence. The available cross-sectional design cannot establish that interpretation simply by using a structural-model vocabulary. The limitations section appropriately asks for longitudinal or experimental research to investigate directionality. [§5.2].
Is the study still useful?
Yes, for a bounded question about reported experience and credibility among these respondents. The detailed instrument lets you see what people were asked, and the sample table preserves their institutional and role context. It is useful to know that apparently technical construct names can contain questions about trust and preferences.
For a B2B company, that can sharpen a customer-research plan. Ask about navigability or clarity when perceptions are the outcome, observe task completion when success is the outcome, and capture actual answers when the question is whether a vendor appeared. Calling each result SEO performance would conceal what was measured.
There is also a trade-off in survey design. Self-report makes it possible to collect experiences that a technical test cannot see. It leaves some technical claims dependent on what respondents can know and remember. That trade-off deserves explanation rather than an automatic dismissal of the survey or an automatic promotion of its coefficients into causal facts.
What belongs beside it from the Index?
Our captured-answer record is a different kind of evidence. Microsoft Dynamics 365 is named in 36/150 valid September CRM target answers, 24.0%, descriptive 95% Wilson interval 17.9% to 31.4%. The numerator comes from a published answer record under a defined matching rule, not a respondent’s rating of the company. Absence Index release 2026-09, CRM vendor table.
That does not make the Index a test of this paper’s engagement model. It supplies a contrast in measurement: a saved answer can support a naming verdict, while a questionnaire can support a statement about reported perception. Neither design supplies the other’s missing outcome.
We grade the Jordan study B for the observational perception evidence used here. The unresolved question is how those attitudes relate to independently observed rankings, repeated answer inclusion or actual decisions under a design that can separate direction and competing explanations.
What our own data shows
Microsoft Dynamics 365 is named in 36/150 valid September CRM target answers, 24.0% (Wilson 95% interval 17.9% to 31.4%). This answer-level measure is separate from the Jordan survey. The September release supplies the independently inspectable naming observation beside this survey. Vendor-level counts and intervals preserve its valid-answer base. We have no measurement here linking those answers to the Jordan respondents’ engagement scores, or estimating a causal effect of engagement on search position. The outside model is reported with its self-report and numerical-reporting limits.
Sources
- Bataineh, A. Q., Qasim, D., Idris, M., & Abu-AlSondos, I. A. (2025). The evolution of SEO strategies: From keywords to user experience in private universities of Jordan. Cogent Business & Management, 12(1), 2491678. https://doi.org/10.1080/23311975.2025.2491678 . Methods §§3.1 to 3.3, Tables 1, 6 and 7, limitations §5.2, Appendix A. Accessible original Verified 30 October 2026.
- Broadcastwell. (2026). Absence Index, release 2026-09 [Data set]. Zenodo. Release JSON, answer JSONL and README. https://doi.org/10.5281/zenodo.22907695 Verified 30 October 2026.
Cite this page
Broadcastwell research team. (2026, October 30). Bataineh and colleagues measure perceptions of university websites. Broadcastwell. https://docs.broadcastwell.com/research/sources/bataineh-2025/ Published 30 October 2026; updated 30 October 2026.
@online{broadcastwell2026research15,
author = {{Broadcastwell research team}},
title = {Bataineh and colleagues measure perceptions of university websites},
year = {2026},
date = {2026-10-30},
url = {https://docs.broadcastwell.com/research/sources/bataineh-2025/},
urldate = {2026-10-30},
note = {Published 2026-10-30; updated 2026-10-30}
}
Prepared by the Broadcastwell research team. Lead: Arjun Mehta, management consultant.
Sources checked 30 October 2026. Independent check: 30 October 2026, Broadcastwell verification desk. This page has not been peer reviewed.
Last updated: 30 October 2026. Next review: 30 January 2027.