Back to Top

 Skip navigation

Sampling

To the best of the Consortium’s knowledge, Ireland followed the PIAAC Technical Standards and Guidelines (TSG) related to sampling and weighting.

  • Sampling plan: Pass
  • Sample selection:
    • Home office: Pass
    • In field: Pass
  • Sample weighting: Pass
  • Sampling error: The effective sample size – because of clustering, differential sampling rates and weighting adjustments, and measurement error – is at an adequate level (greater than 1,000) for proficiency measures.

Coverage and Nonresponse Bias

To the best of the Consortium’s knowledge, Ireland followed the PIAAC Technical Standards and Guidelines (TSG) related to population coverage and nonresponse bias.

  • Population coverage:
    • Frame: Pass The estimated percentage of the target population excluded from the frame was 2% (households on the islands surrounding Ireland)
    • Data collection: Weighted percentage of cases excluded = Not applicable
  • Weighted response rate and coverage weight: Weighted overall BQ response rate (ORR1) = 48.3%; Weighted overall assessment response rate (ORR2)  = 47.0%; Weighted coverage rate  =  9%
  • Nonresponse bias analysis:
    • Basic: Pass
  • Extended: Caution-Low Ireland had an overall BQ response rate (ORR1) of 48.3%. However, the EVO, which can be considered an adjusted response rate that takes into account the effectiveness of the nonresponse and raking adjustments, was 59.6% (for literacy). Ireland did not use adaptive survey design to try to reduce nonresponse bias.

The comparison to external totals showed no notable significant differences in the distribution of the variable that was used in the analysis (region by education level). An analysis using alternative raking dimensions (10-year age group by employment status) showed no notable significant differences in mean scores, therefore this analysis does not show any indication of the potential for bias.

Considering all of the analyses together, and particularly noting that ORR1 and EVO fall below 70%, the conclusion is that users should be cautioned that the potential for nonresponse bias exists in this country's data. Relative to other countries that received a caution, there is less evidence of potential bias.

Data Collection

To the best of the Consortium’s knowledge, Ireland followed the PIAAC Technical Standards and Guidelines (TSG) related to data collection.

  • Staffing: The target number of interviewers planned was 50 as documented in the National Survey Design and Planning Report (NSDPR). Of the 64 trainees who completed training, 63 were assigned to fieldwork. Data collection was successfully completed with the extension to the Batch 2 deadline for data submission.
  • Interviewer Training: Pass
  • Monitoring and Validation: Pass Ireland partially met the reduced requirement for validation. For the purpose of data evaluation, countries were considered to have met the standard if they had validated at least 7% of cases (selected randomly across all dispositions) for 90% of the interviewers. Ireland met the 7% threshold for 87% of the interviewers. Thirteen percent of the interviewers were validated at less than the 7% level. Ireland met the 10% validation requirement for 78% of interviewers; in addition, at least 10% of cases were validated overall. All final disposition codes were validated. Data quality of completed work was generally monitored according to the TSG requirements.

Instrument Data Quality

Translation

To the best of the Consortium’s knowledge, Ireland followed the PIAAC Technical Standards and Guidelines (TSG) associated with translation and verification, in particular: Standard 6.1 for new cognitive items, Standard 6.2 for background questionnaire materials, and Standard 6.3 on trend cognitive items. All adaptations were documented, and all materials went through full verification1 prior to the Field Test and a partial verificationprior to the Main Survey.

  • Outcome: Pass

1Full verification was a sentence-by-sentence check for equivalence to source + linguistic correctness + appropriate/approved adaptations, with a final check that crucial issues identified during verification have been correctly addressed in pre-final instruments.

2Partial verification is a check of correct echoing of FT to MS changes in source version + vetting and verification of other changes at the initiative of countries, with again a final check whereby any remaining issues were addressed and corrected.

Assessment Data

Across all countries, 98.2% of respondents who completed the background questionnaire (BQ) went on to take some of the cognitive assessment. In Ireland, 99% of the respondents who completed the BQ took some of the cognitive assessment.

Across all countries, 96.2% of respondents who had no smartphone, tablet, or laptop/desktop computer experience went on to take some of the cognitive assessment. In Ireland, the percentage of respondents with no computer or device experience who went on to take some of the cognitive assessment was 97.1%. Specifically, 96.2% of respondents with no tablet experience across all countries and 97.4% in Ireland went on to take some of the cognitive assessment.

The assignment of cognitive domains within the PIAAC Delivery System (PDS) accurately followed the intended workflow. That is to say, the administration of Components, Literacy, Numeracy, and APS followed the assessment design and the adaptive routing for Stages 2 and 3 for Literacy and Numeracy were accurately implemented. Analyses also showed accurate data capture for all countries.

  • Cognitive Assessment Overall: Pass

Coding

To the best of the Consortium’s knowledge, Ireland followed the PIAAC Technical Standards and Guidelines (TSG) associated with coding, in particular Standard 11.1.3: 50% of the occupation and sector of industry data will receive a second (quality control) code by a (different) human coder. The reliability target is 85% agreement at the three-digit level.

  • Coding Overall: Pass
  • Double Coding: Standard Met
  • Double Coding by Different Coders: Standard Met
  • Reliability Target of 85% at the three-digit levels: Agreement standard of 85% reliability was not met for 4/6 coded items: ISCO08_C = 77%, ISCO08_L = 79%, ISIC4_C = 75%, ISIC4_L = 78%

Background Questionnaire Data

Across all countries, 99.4% of respondents who started the BQ completed it. In Ireland, 99.82% of the respondents who started the BQ completed it. Because of this high completion rate, individual items in the BQ had a very high response rate as well. There was one exception: Income related questions. Across all countries, 94.6% of respondents reported their income either directly (88.2%) or in broad categories (6.4%). In Ireland, 96.8% of respondents provided their income either by reporting the direct amount (92.2%) or in broad categories (4.6%).

If a respondent decided to break off the interview, the interviewer was able to collect a reason for the breakoff. The data contains about 0.7% of cases with breakoff codes across all countries. In Ireland, we observed 0.18% of cases with breakoffs.

Across all countries, 1.7% of respondents completed the Doorstep Interview due to a language barrier in the BQ language(s) offered. In Ireland, the percentage of Doorstep respondents was 0.29%.

  • Background Questionnaire: Pass

Item Nonresponse

Across all countries, for computer-based items, the average proportions of nonresponse (omitted or not reached) was 13.2% for Literacy, 11.1% for Numeracy, and 12.1% for APS. In Ireland, the percentage of nonresponse for Literacy was 11.1%, for Numeracy it was 9.2%, and for APS it was 10.8%.

On average, countries had 9.9 items in the BQ that had a response rate less than 85%. Ireland had 4 items with a response rate less than 85%. Note that there are over 300 individual items in the PIAAC Cycle 2 BQ. Like in Cycle 1, valid skips were determined based on international routings. National BQ adaptations can include minor changes to routing to accommodate necessary item adaptations. Because these national routing adaptations are not taken into account when calculating valid skips, some countries may have an exaggerated rate of nonresponse for some items.