Qualitative versus quantitative forecasting
A planning director must forecast demand for a product about to launch or one sold for years: should the forecast rest on expert judgement or on historical data?
- Origin
- Hyndman and Athanasopoulos, 2021; Dalkey and Helmer, 1963
- Diagram type
- Comparison table
- Level of analysis
- Firm
- Topics
- Data & analytics
Diagram
Qualitative versus quantitative forecasting
| Criterion | Qualitative (judgmental) | Quantitative (statistical) |
|---|---|---|
| Use when | No data, or data not relevant: a new product, a new competitor, market conditions with no historical precedent | Numerical information about the past exists and it is reasonable to assume some past patterns will continue |
| Techniques | Delphi, forecasting by analogy, scenarios, sales force composite, executive opinion, customer intentions surveys | Time series models (exponential smoothing, ARIMA) and explanatory models such as regression |
| Weakness | Subjective: inconsistent, anchored on familiar reference points, prone to optimism | Only sound while the future resembles the past in the patterns that were modelled |
| Principle | Make it structured: set the task clearly, document the reasoning, separate forecasters from users | When data exist, start there; statistical forecasts are generally better than judgement alone |
When to use it, and when not to
Use it when
New product or no data: use structured judgement such as Delphi, analogy or scenarios. Data available: start with a statistical model and adjust by judgement only when there is significant extra information the model lacks.
Pitfalls
Judgement carries bias: salespeople who forecast and are also judged on sales may push forecasts up or down, and group meetings feed optimism. Delphi curbs this with anonymous experts but takes time: two or three rounds are usually enough, and more rounds mean more drop-outs.
Sources
References
Listed A–Z by author surname, Harvard style. A source with a DOI links to https://doi.org/…; one without links to its page or says plainly that it has no DOI.
- Dalkey, N. and Helmer, O. (1963) ‘An experimental application of the Delphi method to the use of experts’, Management Science, 9(3), pp. 458–467.doi:10.1287/mnsc.9.3.458Source page
- Hyndman, R.J. and Athanasopoulos, G. (2021) Forecasting: principles and practice (3rd edn). Melbourne: OTexts.No DOIotexts.comCited in 2 articles