Wêrom de Replikaasjekrisis it Fertrouwen yn Wittenskip Skeind Hat
June 4, 2026 · Frisian News
More than half of published scientific studies cannot be reproduced, revealing systemic failures in how research gets funded, conducted, and rewarded. The crisis extends beyond labs into policy decisions that affect millions of people.
FryskYn 2015 besochten ûndersikers fan it Center for Open Science 100 psychologyske ûndersiken te reprodusearjen dy't yn grutte tydskriften publisearre wiene. Sy slaggen mar yn 36 gefallen. Dit oantal ferbjustere net folle fan binnenút, want in protte wittenskippers wisten al dat it systeem stikken wie. It probleem giet djipper as sloardigens. Tydskriften jouwe de foarkar oan ferbaasgjende finen boppe krekte. Ûndersikers dy't dramatyske resultaten publisearje krije beurzen, promoasjes en sitaten. Dejingen dy't negative finen rapportearje sammelje stof. De prikkelstruktuer beleant flaters.
Grutte farmabedriuwen ferergerje it probleem. In farmabedriuw betellet foar in stúdzje dy't oantoant dat syn produkt wurket. Wittenskippers dy't troch dat bedriuw yn tsjinst naam wurde liede it ûndersyk. It bedriuw kontrôlearret hokker resultaten publisearre wurde. Stúdzjes mei ûngeunstiche útkomsten ferdwine yn arsjyfkasten. In analyse út 2023 fûn dat troch fabrikanten betelle ûndersiken nei genêsmiddelen positive resultaten twa kear sa faak rapportearje as ûnôfhinklike ûndersiken dy't deselde genêsmiddelen teste. It publyk sjocht de mislearre eksperiminten nea. De autoriteiten karre genêsmiddelen goed op basis fan ûnfolsleine gegevens.
Universiteiten twinge ûndersikers nei produktiviteitsmetriken dy't neat as folume mitte. Publisearje 50 papers yn it jier en do klimste de ljedder op. Besteegje fiif jier oan it goed krijen fan ien resultaat en do rekkest efterút. Jonge wittenskippers leare gau: snij hoeken ôf, fyn statistyske trúkjes, massearje de gegevens oant de p-wearde ûnder de 0,05 falt en it tydskrift it akseptearret. Peer review mislearret yn it opfangen fan it meastepart hjirfan om't reviewers ûnbetelle wurkje, tsientallen papers yn 'e moanne beoardielje en gjin tiid of motivaasje hawwe om berekkeningen te kontrôlearjen of rauwe gegevens op te easkjen.
De kosten falle it hurdst op gewoane minsken. In stúdzje wurdt publisearre dy't stelt dat in fiedingssupplement hertsikte foarkomt. Mediakanalen fergrutsje de claim. It supplement ferkeapet miljoenen flessen. Trije jier letter besykje in goed finansiearre team it te werheljen en fine neat. Tsjin dy tiid is de skea oanrjochte. Rjochtlinen foar folksgesûnens ferskowe op basis fan de earste flashy fining yn plak fan kumulatyf bewiis. Regearingen ferspille jild oan yntervinsjes dy't nea yn earsten wurken. Fertrouwen eroadearret net om't wittenskip ynherint korrupt is, mar om't it systeem wierheid as sekundêr behannelet foar karriêrefoarútgong.
Guon tydskriften easkje no dat ûndersikers ûndersiken registrearje foardat sy se útfiere, har fan tefoaren oan har analyseplan bine en rauwe gegevens iepenbier diele. Dizze feroarings wurkje. Registrearre ûndersiken toane legere persintaazjes ûnmooglike resultaten. Mar de measte tydskriften negearje dizze noarmen. De measte ûndersikers mije se. De kulturele feroaring fereasket dat wy erkenne dat it hjoeddeiske systeem mislearre, en ynstellings kontrôlearje harsels selden sa earlik.
In 2015, researchers at the Center for Open Science tried to reproduce 100 psychological studies published in major journals. They succeeded in only 36 cases. That number shocked few insiders because many scientists already knew the system was broken. The problem runs deeper than careless mistakes. Journals prefer surprising findings over accurate ones. Researchers who publish dramatic results get grants, promotions, and citations. Those who report negative findings gather dust. The incentive structure rewards error.
Big pharma compounds the problem. A drug company pays for a study showing its product works. Scientists employed by that company lead the research. The company controls which results get published. Studies with unfavorable outcomes vanish into filing cabinets. A 2023 analysis found that drug trials sponsored by manufacturers report positive results at twice the rate of independent trials testing the same drugs. The public never sees the failed experiments. Regulators approve medicines based on incomplete evidence.
Universities push researchers toward productivity metrics that measure nothing except volume. Publish 50 papers a year and you climb the ladder. Spend five years getting one result right and you fall behind. Junior scientists learn fast: cut corners, find statistical tricks, massage the data until the p-value drops below 0.05 and the journal accepts it. Peer review fails to catch most of this because reviewers work unpaid, review dozens of papers monthly, and lack the time or motivation to check calculations or demand raw data.
The cost falls hardest on ordinary people. A study gets published claiming a diet supplement prevents heart disease. Media outlets amplify the claim. The supplement sells millions of bottles. Three years later, a well-funded team attempts replication and finds nothing. By then the damage is done. Public health guidance shifts based on the first flashy finding rather than cumulative evidence. Governments waste money on interventions that never worked in the first place. Trust erodes not because science is inherently corrupt but because the system treats truth as secondary to career advancement.
Some journals now require researchers to register studies before running them, precommit to their analysis plan, and share raw data publicly. These changes work. Registered studies show lower rates of impossible results. But most journals ignore these standards. Most researchers avoid them. The cultural change requires admitting the current system failed, and institutions rarely police themselves that honestly.
Published June 4, 2026 · Frisian News · Ljouwert, Fryslân